SlideShare a Scribd company logo
*Financial support for this research was provided by the Listed Company Association,
Thailand. The authors appreciate comments and suggestions by four anonymous referees and
participants at presentationsat theThird AnnualEuropean FMAConference (Barcelona 1999), the
Sixth Annual Conference of the Multinational Finance Society (Toronto 1999), and the Eleventh
Annual PACAP/FMA Finance Conference (Singapore 1999).
(Multinational Finance Journal, 1999, vol. 3, no. 2, pp. 103–125)
© Multinational Finance Society, a nonprofit corporation. All rights reserved.
1
An Investigation of Thai Listed Firms’
Financial Distress Using Macro and Micro
Variables*
Sunti Tirapat
Chulalongkorn University, Thailand
Aekkachai Nittayagasetwat
National Institute of Development Administration, Thailand
The emergence of the economic crisis in Thailand in 1997 is an interesting
case for academic studies. Internationally, it had a contagion effect, spreading
to countries in Asia and in other regions. Domestically, it caused a great many
industrial and corporate bankruptcies. The Thai economy had been relatively
stable since 1984. The recent development in 1997, however, produced a
sudden economic slumpresultinginclosuresof manyThai corporations. Using
alogitregression,thisstudydevelopsamacro-relatedmicro-crisisinvestigation
model. The significance of the model is in its ability to bridge a firm’s
sensitivity to macroeconomic conditions and its financial characteristics in
order to explore a firm’s financial distress. The findings indicate that
macroeconomicconditionsarecriticalindicatorsof potentialfinancialcrisisfor
a firm. The article shows that the higher a firm’s sensitivity to inflation, the
higher the firm’s exposure to financial distress.
Keywords: bankruptcy, financial distress, prediction model, and Thailand
crisis.
I. Introduction
In 1997, an Asian financial crisis emerged. From Thailand, the crisis
rapidly spread throughout the region. The swiftness and the
confounding effects of the “Tom Yam Kung” crisis have raised several
Multinational Finance Journal104
1. “Tom Yam Kung” is a famous Thai dish and it became the name of the 1997
economic crisis spreading from Thailand.
2. The speculative attack literature is reviewed by Flood and Marion (1998). The
contagious nature of the crisis has been investigated by Eichengreen, Rose, and Wyplosz
(1996) and more recently by Glick and Rose (1998).
3. During 1992-1995, the growth of average net incomes of firms listed in the Stock
Exchange of Thailand was around 22% while in the 1996 it declined to – 31.3%.
4. Studies that use only financial ratios include early works such as Beaver (1966) and
Altman (1968). Much of the recent work is discussed in Altman and Saunders (1998).
Studies that incorporate accountant opinions include Hopwood et al. (1989) and Flagg et al.
(1991). Studies that use only financial ratios include early works such as Beaver (1966) and
Altman (1968), and Altman, Haldeman, and Narayanan (1977). Much of the recent work is
discussed in Altman and Saunders (1998) and Zavgren (1983). Studies that incorporate
accountant opinions include Hopwood et al. (1989), Flagg et al. (1991).
5. For example, in Thailand fewer than 10 companies listed in the Stock Exchange of
Thailand were financially distressed during 1987-1994.
interesting economic questions.1
What was the root of the problem?
Why was the crisis ‘contagious’? What triggered the speculative
attack? These issues have already been investigated by prominent
economic scholars. Numerous causes of the crisis have been proposed,
ranging from the role of financial panic (Rachel and Sachs [1998]) to a
nexus of moral hazard problems (Corsetti, Pensenti, and Roubini
[1998]).2
The crisis, however, has provided an opportunity to investigate
another area of interest, the role of corporate bankruptcy. The impact
of the crisis has resulted in a large number of Thai firms going bankrupt
in 1997.3
Since the massive bankruptcy is an exceptional case in
Thailand, this article aims to seek the cause of this event. Prior articles
on predicting bankruptcy focus primarily on financial variables such as
financial ratios, cash flows, or accountingopinions.4
Many studies have
been conducted in developed countries because data are readily
available. In emerging markets, however, this line of research is scarce
since, in growing economies, bankrupt firms are actually rare.5
The
crisis provides us with an opportunity to investigate the prediction
model in developing countries such as Thailand. An investigation of the
financial distress model may be of interest and provide some insights
into this area of research since the Thai environment may differ from
that of the developed markets on several aspects. First, the institutional
structure of the financial sector in Thailand can be characterized as a
bank-centered financial system, in contrast with that of a market-based
Investigation of Financial Distress for Thai Firms 105
6. See La porta, Lopez-de-Silanes, Shleifer, and Vishy (1998), which compile
ownership concentration of 45 developed and developing countries.
7. Alba,Claessens,andDjankov(1998)pointoutthataccountingandauditingpractices
in Thailand may be due to the lack of strong self-regulatory organizations in the auditing and
accounting profession and a shortage of well-qualified accountants and auditors.
system such as that in the U.S. The credits from financial institutions
(commercial banks and finance companies) are about 5.5 times larger
than funds raised (in form of equity and bonds) from capital markets in
1994 and 16.7 times larger than funds raised in 1997. Secondly, the
ownership structure in Thailand is highly concentrated. The average
percentage of common shares owned by the three largest shareholders
in the ten largest firms is around 44 percent.6
The typical Thai
conglomerate is controlled by family and highly leveraged by local
financial institutions. This structure has led to inefficient investments
before the 1997 financial crisis. The ratio of earnings before interest and
taxes (EBIT) over interest expenses declined from 5.78 in 1994 to 1.49
in 1997, while the leverage ratio rose from 1.5 to 2.95 during the same
period. Thirdly, anecdotal evidence shows that although the accounting
standards in Thailand are similar to those of other international
standards such as the International Accounting Standard (IAS) or the
Financial AccountingStandardBoard(FASB),theaccountingpractices
are of concern.7
Hence, it is interesting to see whether or not financial
ratios can appropriately explain the causes of failure of firms in
developing countries such as Thailand.
In addition, an investigation of financially distressed firms in
emerging markets, in our view, should take economic conditions into
account since these firms are typically small and operating in an open
economy. Moreover, as a firm’s accounting system is subject to
accounting standard, e.g., Generally Accepted Accounting Principles
(GAAP), it seldom reflects immediate changes in a firm’s financial
situation when a crisis strikes the overall economy. Therefore,
accountingvariablesalonemaynotadequatelyexplainafirm’sfinancial
distress, and the firm’s sensitivity to macroeconomic factors should be
taken as a significant explanatory variable to indicate a firm’s financial
distress. In other words, firms that are more sensitive to economic
shocks should be more prone to financial distress than firms that are less
sensitive. This article attempts to incorporate the macroeconomic
factors in the investigation of financially distressed firms. The purpose
of incorporating macroeconomic variables is to capture the sensitivity
Multinational Finance Journal106
to change in macroeconomic factors. This perspective, in our view, is
more consistent with the Thai situation. Using the two-stop logit
approach,thisstudydevelops a macro-relatedmicro-crisisinvestigation
model. It is found that the sensitivity of firms to economic conditions
playsamajorroleindifferentiatingthe financiallydistressedcompanies
from the non-distressed ones.
The article is organized as follows: Section IIreviews prior research.
The sample and methodology are discussed in section III. Empirical
results of the model are reported in section IV. The discussion of the
findings and conclusions are presented in section V.
II. Literature Review
A. Prediction of Bankruptcy
Bankruptcy prediction models were pioneered by Beaver’s (1966)
univariate test and Altman’s (1968) multivariate discriminant analysis.
Both articles document that financial variables can be used to predict
bankruptcy. Since then, prediction of corporate failure has been a topic
of much interest and more recent works have extended this line of
research in three areas: statistical techniques, definitions of bankruptcy,
and a greater variety of explanatory variables. For example, Ohlson
(1980) utilized the logit and probit analysis to estimate the probability
of bankruptcy. The adjustments of estimation bias resulting from
oversampling, such as weights based on prior probabilities in the
estimationprocessandoptimalcut-offpoint,arediscussedinZmijewski
(1984). The problems of pooling data due to a small number of samples
are addressed in Zavgren (1983) .
The second area of extension has dealt with the definitions of
bankruptcy. For example, a model that distinguishes between
financially distressed firms that survive and financially distressed firms
that ultimately go bankrupt has been investigated. Gilbert, Menon, and
Schwartz (1990) find different explanatory financial variables for these
two groups of firms. In addition, a sample firm may be classified into
more than two categories (bankrupt or nonbankrupt) and the
classification probabilities can be estimated by the multinomial logit
technique. For example, Poston, Harmon, and Gramlich (1994)
assigned firms into one of three groups according to each firm's
financial condition: turnarounds, business failures, and survivors. They
Investigation of Financial Distress for Thai Firms 107
8. The specification of their models includes four financial ratios, sales growth, and the
interaction of industry output and financial ratios.
9. Caprio and Klingebiel (1996) document 49 banking crises in the 1970s and 1980s;
and 33 crises occurred over the first six years of the present decade.
find that financial ratios are not so useful in distinguishing between
financially distressed firms that are able to turn around and those that
are unable to avoid failure. Johnsen and Melicher (1994), on the other
hand, suggest that by using multinomial logit models, the classification
errors can be significantly reduced.
The third area involves some adjustment of explanatory variables
either by covering additional variables other than financial ratios or
industry-adjusted ratios. For example, Hopwood, McKeown, and
Mutchler (1989) and Flagg, Giroux, and Wiggins (1991) find that a
qualified opinion is significant in distinguishing financially distressed
firms. Some studies include macroeconomic variables to control for
changes in the business environment. Rose and Giroux (1982) examine
28 business cycle indicators and find that economic conditions affect
the failure process. Mensah (1983) evaluates the bankruptcy model
usingtheprice-leveladjusted (SPL) data. ThefindingsindicatethatSPL
data do not significantly improve bankruptcy prediction. Finally, Platt
and Platt (1990) control for industry differences by using industry-
normalizing ratios. The industry-relative framework results in stable
bankruptcymodels. Therefore,industrialgrowthhasasignificanteffect
on corporate failure. Furthermore, Platt and Platt (1991) investigate the
stability and completeness of a bankruptcy model based on industry-
relative ratios compared to that of unadjusted ratios.8
Both models yield stable parameter estimation. The industry-
relative specification, however, appears to provide incremental
information not contained in the model based on the unadjusted ratios.
B. Prediction of Currency and Banking Crises
Another strand of related research concerns the prediction of currency
and banking crises; for example, Demirguc-Kunt and Detragiache
(1997), Eichengreen and Rose (1998). These studies are motivated by
the recent spate of crises9
in Europe (Finland, Sweden), Latin America
(Mexico, Argentina) and Asia (Thailand, Indonesia). Kaminsky and
Reinhart (1996), in analyzing a sample of 20 crises, report the
importance of macroeconomic factors in a crisis. They document that
Multinational Finance Journal108
output, the stock market, and the real exchange rate usually peak about
a year before the onset of a banking crisis. Real interest rates and bank
deposits rise in the period leading up to the crisis. Demirguc-Kunt and
Detragiache (1997) applya multivariate logit model to the determinants
of banking in a sample of developed as well as developing countries
between 1980 and 1994. They find that in general, low growth and high
inflation are associated with the probability of a banking crisis.
EichengreenandRose(1998)restricttheir sample to cover onlybanking
crises in emerging markets from 1975 through 1992. They apply a
multivariate probit model to nine economic variables; four of their
variables are “international” in nature,three are “domestic”, and two are
“external” in nature. Their results suggest that banking crises in
emerging countries tend to occur in response to a conjuncture of
unfavorable developments in international markets and domestic
imbalances. Specifically, rising industrial-country interest rates
precipitate banking crises: a one-percent increase in interest rates is
associated with an increase in the probability of banking crises of
around three percent.
Kaminsky, Lizondo, and Reinhart (1997) examine the empirical
evidence on 25 currency crises and compare the merits of alternative
approaches in providing early indications of the crises. They document
that the best track record includes exports, deviations of the real
exchange rate from the trend, the ratio of broad money supply to gross
international reserves, output, and equity prices. Kaminsky and
Reinhart (1996) also investigate 76 currency crises and 26 banking
crises and find that banking and currency crises are closely related in the
aftermath of financial liberalization, with banking crises generally
preceding currency crashes.
III. Data and Methodology
A. Sample and Data
The sample includes firms listed in the Stock Exchange of Thailand
(SET)thatexperiencedfinancialdistressin1997. Financiallydistressed
firms are defined in this article as firms that either (1) were closed down
by governmental authorities (all of which were banks and finance
companies) or (2) were required by the Bank of Thailand or SET to
submit restructuring plans (these companies were designated as C or
Investigation of Financial Distress for Thai Firms 109
10. The SET posts a supervision sign of “C” (the filing for compliance) or “SP*” (the
suspensionor thetemporaryprohibitionof tradinguntil thecauses of delisting areeliminated)
against a listed company and the firm must submit the SET documents supporting the
company’s financial position.
SP10
by the Exchange). There were 459 firms listed in the SET in 1996,
of which 55 were financially distressed by this definition. Firms that
were first listed on the SET in 1995 or later and firms with incomplete
data are excluded. The resulting sample consists of 341 non-financially
distressed firms and 55 financially distressed firms. The characteristics
of firms in the sample are reported in table 1.
B. Method
Although some prior research, e.g., Rose, Andrews, and Giroux (1982),
have included macroeconomic variables in the prediction model, this
study differs on one important aspect. The purpose of incorporating
macroeconomic variables is not to control for changes in the business
environment but rather to capture the sensitivity to change in
macroeconomic factors. This perspective, in our view, is more
consistent with the Thai situation.
The sensitivity of firms to economic conditions is expected to play
a major role in differentiatingthe financiallydistressed companies from
the non-distressed ones. For example, an enterprise, which has a high
systematic risk corresponding to macroeconomic conditions, e.g.,
growth of industrial production, inflation, changes in interest rates, and
changes in money supply (M2), may be exposed to the possibility of
financial distress. In this article, the multi-factor model represents the
relationship between the firm’s stock return and the firm’s sensitivity to
macroeconomic factors. In other words, changes in macroeconomic
factors and the firm’s sensitivity to those changes will affect the firm’s
stock return. Then, the relationship between the firm’s probability of
financial distress and the firm’s stock return is such that a firm, which
encounters a high return, will have a low probability of financial
distress. This process needs a two-step analysis whereby changes in
macroeconomic factors and the firm’s sensitivity to those factors affect
the firm’s stock return, with a firm’s stock return in turn having an
effect on the firm’s probabilityof financial distress. Therefore, the two-
step logit can be applied as used in Maddala (1986). This line of
methodology has also been used in Thomson (1992) and Theodossiou
et al. (1996). For example, the study by Thomson (1992) uses two-step
MultinationalFinanceJournal110
TABLE 1. Sample Characteristics
Description Full Sample Distressed Non-distressed
Total assets 19,772,150.10 21,690,440.67 19,462,748.39
Earnings before interest and taxes 512,615.58 –84,358.06 608,901.65
Total equity 3,371,051.44 2,683,971.35 3,481,870.81
Total debt 16,401,098.66 19,006,469.32 15,980,877.58
Number of observations 396 55 341
Note: The figures are expressed in thousand baht. As of December 31, 1996, the exchange rate between the US$ and the Thai baht is US$ 1 = 25.7
baht. However, after the Ministry of Finance of Thailand changed the exchange rate regime from the currency basket to the managed float system on
July 2, 1997, the currency exchange of US$ 1 as of December 31, 1997, was equal to 48.15 baht.
Investigation of Financial Distress for Thai Firms 111
logit regression approach as a link between official bank failure and
bank insolvency. Theodossiou et al (1996) also reexamine the
economic factors determining the decision to acquire a financially
distressed firm using a sequential response logit (SRL) model. In their
model,acquisition,nonacquisition,andwealthyoutcomesareexpressed
as a two-stage sequence of binary outcomes, healthy/distress in stage
one, and acquisition/non-acquisition in stage two.
Therefore, logistic discriminant analysis used in this article will
apply this two-step idea in order to estimate return on a firm’s stock,
and then the estimated stock return will be used as a representation of
a macroeconomic sensitivity indicating a firm’s financial distress. The
estimated stock return calculated from the multi-factor model is
designed as a link to the micro-crisis model. In other words, the firm’s
estimated return calculated from the multi-factor (or macroeconomic
factors) model represents the macro effects incorporated into the micro-
crisis prediction model.
Several researchers, e.g., Beaver (1966) and Scott (1981), confirm
the relationship between a firm’s probability of bankruptcy and the
firm’s stock return. But because the firm’s actual stock return reflects
both the systematic risk and firm-specific risk, the motivation of the
two-step is to capture only the systematic risk in which the economic
crisis has affected the firms in the economy as a whole. In general, the
micro-crisis model and the related macro factors imply the following
specifications (the indirect test):
,prob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
where
, (1),
ˆ
i i j i
j
Z a bR c= + + ∑
,$ $
, ,R Fi k i k
k
= + ∑β β0
Yi
is assigned the value of 1 if firm i is financially distressed and the
value of zero otherwise, Prob(Yi=1) is the probability with which firm
i is classified as a financially distressed firm, Zi is a linear function
in which a, b, and c are coefficients of variables, Xj,i is the financial
Multinational Finance Journal112
characteristic j of firm i, is the estimated macroeconomic factor k,$Fk
and is the estimated monthly stock return of firm i, specified as a$Ri
linear function of the macroeconomic factor .$Fk
The probability, Prob (Yi=1), depends on the logistic function of a
linear relationship of a set of variables including and Xi,j. The$Ri
estimated monthly stock return of the firm i, , can be calculated from$Ri
amulti-factormodelinwhichtheestimatedmacroeconomicfactors, ,$Fk
are used to develop the model. The relationship between a firm’s
probability of financial distress and its stock return has been pointed out
by Beaver (1968) and Scott (1981). While the first study showed the
decline of a firm’s stock return as the firm approached failure, the
second confirmed that the probability of bankruptcy depended on the
firm’s stock return. Therefore, a firm’s stock return is a good reflection
of the market expectation of the firm’s probability of financial distress.
The actual stock return, Ri , however, represents both systematic risk
andfirm-specificrisk;e.g., ,inwhichFrepresentsR F ei i i i= + +β β0 1, ,
the systematic factor(s) and ei contains the firm-specific risk. Since the
firm’s financial characteristics, discussed later, are related to the firm-
specific risk, the use of the actual stock return, together with the firm’s
financial characteristics, as explanatory variables to the model will be
a double consideration of the firm-specific risk. In addition, the 1997
economic crisis has systematically hit the overall listed firms, so the
motivation of this study is an attempt to find a proxy for systematic risk
corresponding to the firm’s bankruptcy risk. The estimation
specification of will carve out the firm-specific risk and capture only$Ri
the systematic risk of the firm. Then the use and financial$Ri
characteristics combines the systematic and firm-specific risks as the
representationofexplanatoryvariablestothemacro-relatedmicro-crisis
model without a redundancy of the firm-specific risk consideration.
C. Financial variables
The set of variables, Xj,i, represents financial characteristics of the firm
i. Shivaswamy,Hoban,andMatsumoto(1993)studiedthirteenresearch
papers and summarized the frequency of financial ratios used in those
papers. The most frequently used ratios were: current ratio, leverage
ratio, and profitability ratio. Altman and Altman et al. (1968, 1983,
1984, 1994) developed numerous models predicting bankruptcy and
financial distress. In particular, Altman (1984) surveyed studies on
Investigation of Financial Distress for Thai Firms 113
businessfailuremodelsinvarious countries and reviewedtheutilization
of variable profiles employed in those studies. The most frequently
used financial ratios were: working capital to total assets, retained
earnings to total assets, earnings before interest and taxes to total assets,
market value of equity to total liabilities, and sales to total assets.
However, the literature in this area of finance and accounting shows
that researchers usually base their models on the capital, assets,
management, earnings, and liquidity of a firm, which are the so-called
CAMEL categories. FollowingSalchenberger, Cinar, andLash (1992),
the financial ratios are categorized only from each element of the
CAMEL framework. The set of variables, Xj, following the CAMEL
categories is selected as follows:
Capital: X1 = book value of stockholders’ equity / total assets (SETA)
Assets: X2 = retained earnings / total assets (RETA)
Management and Earnings: X3 = operating income / net sales (OINS)
Liquidity: X4 = net working capital / total assets (WCTA)
These four variables will be used as the financial characteristics of the
firms in the sample and as explanatory factors to the model.
The dependence on a set of financial ratios as a sole explanatory set
of variables, however, has raised a question of whether they are
sufficient to explain a firm’s financial situation when the crisis strikes
the economy as a whole. The reason is that a firm’s accounting system
is prone to accounting standard, e.g., Generally Accepted Accounting
Principles (GAAP), and it may not represent a good rationale for a
firm’s financial distress due to the economic slump. So a firm’s
sensitivitytomacroeconomicfactorsshouldbeincluded,as explanatory
variables to the firm’s financial distress and the estimated stock return
whichincorporatesthe effect of the firm’s sensitivitytomacroeconomic
factors should be a good representation of the macroeconomic factor for
explaining the firm’s probability of financial distress.
D. Multi-factor model
Demirguc-Kunt and Detragiache (1997) use ten variables as
macroeconomic factors in their logit model. However, only four
variables (real quarterly GDP growth, real short-term interest rate,
inflation,andM2moneysupplytointernationalreserves)aresignificant
variables, which can explain the probability of banking crises. These
four macroeconomic factors have also been widely used
Multinational Finance Journal114
econometrically. For the sample of Thai economic conditions, monthly
growth of the production-manufacturing index will provide more
observations than will the use of quarterly real GDP growth. Therefore,
the macro-factors, Fk, are selected as follows:
F1 = monthly growth of the production manufacturing index (PMI)
F2 = monthly inflation or changes in the consumer price index (CPI)
F3 = monthly changes in interest rates (INT)
F4 = monthly changes in M2 money supply (MS2)
The estimated changes in the above economic conditions will be
used to calculate the estimated stock returns of firm i according to the
following equation:
,$ $ $ $ $
, , , , ,R F F F Fi i i i i i= + + + +β β β β β0 1 1 2 2 3 3 4 4
where 0,i is the intercept and 1,i, 2,i, 3,i, and 4,i are coefficients of
variables representing the systematic risks or the sensitivity of firm i to
the changes in industrial production, consumer price index, interest
rates, and M2 money supply, respectively.
It should be noted that for the specification in the indirect test
(equation 1), the estimated stock return is a complete reflection of the
sensitivity of the firm to macroeconomic conditions as well as the
macroeconomic conditions. Alternatively, the effect of a firm’s
sensitivity of macroeconomic factors on the probability of the firm’s
financial distress can be specified as follows (the direct test):
prob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
where (2)
.Z a b c X ei k k i j j i i
jk
= + + +∑∑ β , ,
The above model tests the effect of the sensitivity of firms in terms
of financial distress to four macroeconomic conditions: growth of
industrial production, inflation, changes in interest rates, and changes
in M2 money supply. In this direct test, the explanatory factors of the
firm’s probability of financial distress include the firm’s sensitivity to
macroeconomic factors and the firm’s financial characteristics.
Investigation of Financial Distress for Thai Firms 115
11. The monthly data of the stock return and the macroeconomic factors were collected
startingfromJanuary1987 toDecember1996. Therefore,thereare120monthlyobservations
that are employed in regressing the stock return over the macroeconomic factors.
The systematic risks, k,i, of firm i are obtained from the regression
of the monthly stock returns of the firm i at time ti, Ri,t on the monthly
macroeconomic factors at time ti, Fj,t.11
The equation is such that:
,R F F F F ei i i t i t i t i t i t= + + + + +β β β β β0 1 1 2 2 3 3 4 4, , , , , , , , , ,
where ei,t, the error term, represents the firm-specific risk of the firm i
at time t.
IV. Empirical Results
The descriptive statistics of variables are reported in table 2, while the
resultsofthemacro-relatedmicro-crisispredictionmodels,bothindirect
and direct tests, are presented in table 3. The results indicate that the
estimated stock return variable corresponding to the estimated changes
in macroeconomic factors is significantly different from zero. The
negative sign suggests that a firm with a higher estimated rate of return
has a lower probability of financial distress. The rationale has been
pointed out by Beaver (1966) as the declining of a firm’s stock return
signaling the firm’s vulnerability to the possibility of bankruptcy. The
significance of this variable (at the 5% significance level) also indicates
the importance of macroeconomic conditions and those conditions can
strongly influence the probability of the firm’s financial distress.
The results of the alternative direct specification (equation 2) are
also reported in table 3. The results indicate that the only significant
macro factor is the sensitivity of the firms to inflation or changes in the
consumer price index. In other words, only the systematic risk of a firm
exposed to inflation affects the probability of the firm’s financial
distress. The sign suggests that the more a firm faces inflation risk, the
higher the firm’s vulnerability to the risk of financial distress.
For both alternative models (the indirect test in equation 1 and the
direct test in equation 2), the financial characteristics such as
stockholders’ equity to total assets (SETA) and working capital to total
assets (WCTA) are significant at the 1% significance level. The
relationship is such that the greater the SETA and WCTA ratios, the
MultinationalFinanceJournal116
TABLE 2. Descriptive Statistics of Variables for the Full, the Non-distressed, and the Distressed Samples
Full Sample (396 Firms) Non-distressed Sample (341 Firms) Distressed Sample (55 Firms)
Model Mean
Variable Max Min Mean Max Min Mean Max Min Mean Difference
A. Macroeconomic Factor
Ri 3.5138 –11.5892 –.2962 3.5138 –2.5304 –.2214 .9808 –11.5892 –.7602 .5389
[.76] (2.3)** (4.16)*** {6.43}***
PMI 5.6707 –1.441 .0374 1.6086 –1.441 .0243 5.6707 –.8706 .1185 –.0942
[.48] (–.82) (–1.34) {4.45}***
CPI 26.303 –17.5141 1.1273 17.8946 –17.5141 .6371 26.303 –6.9741 4.166 –3.5288
[5.1] (–3.8)*** (–4.89)*** {2.03}***
int .4366 –1.6489 –.0664 .4366 –.5001 –.0586 .1775 –1.6489 –.1143 .0557
[.14] (1.65) (2.67)*** {4.19}***
MS2 26.6999 –18.5894 1.5073 11.7656 –18.5894 1.314 26.6999 –5.3813 2.706 –1.3921
[3.42] (–2.05)** (–2.82)*** {2.49}***
InvestigationofFinancialDistressforThaiFirms117
TABLE 2. (Continued)
B. Financial Characteristic
SETA 1.003 –2.1488 .4201 1.0030 –2.1488 .4601 .7051 –1.1681 .1718 .2883
[.24] (7.26)*** (7.41)*** {1.06}
RETA 1.3293 –3.0053 .1007 1.3293 –3.0053 .1331 .4259 –2.9265 –.098 .2308
[.30] (3.13)*** (4.6)*** {3.06}***
OINS 7.195 –17.1278 .0893 7.195 –17.1278 .1193 .42 –3.4021 –.0965 .2158
[1.1] (1.83)* (1.29) {2.76}***
WCTA 5.3238 –44.2817 –.405 5.3238 –44.2817 –.2145 .3336 –7.5143 –1.5864 1.3719
[2.43] (4.64)*** (3.86)*** {1.67}**
Note: Ri= 0,i+ 1,iF1,t+ 2,iF2,t+ 3,iF3,t+ 4,iF4,t +ei,t The macroeconomic variables are from monthly data during 1987-1996.
á
i is the estimated monthly
stock return of the firm i. PMI represents the systematic risks or the sensitivity of a firm to the changes in industrial production. CPI represents the
systematic risks or the sensitivity of a firm to the changes in consumer price index. INT represents the systematic risks or the sensitivity of a firm to
the changes in interest rates. MS2 represents the systematic risks or the sensitivity of a firm to the changes in M2 money supply. SETA represents the
ratio of stockholders’equity to total assets. RETA represents the ratio of retained earnings to total assets. OINS represents the ratio of operating income
to net sales. WCTA represents the ratio of working capital to total assets. Parentheses include the standard errors in [ ], t-values, for H0: variances are
equal, in ( ), and F-values, for H0: means are not significantly different from zero, in { }. ***, **, and * denote statistical significance at the .01, .05,
and .1 levels, respectively.
Multinational Finance Journal118
TABLE 3. The Macro-related Micro-crisis Prediction Models: Estimated
Parameters and Wald Chi-square Statistics in Parentheses
Estimates of Estimates of
Coefficients Elasticitya
Coefficients Elasticityb
Constant –.9036 n/a –1.1033 n/a
(8.51)*** (11.32)***
Macroeconomic Factor
–.503 .3014
(5.69)**
PMI
–.1288 .0795
(.12)
CPI
.0986 .4435
(8.7)***
INT
–1.5497 .1695
(1.25)
MS2
–.0211 .0708
(.15)
Financial Characteristic
SETA –3.5065 .7899 –3.1373 .6993
(18.64)*** (14.66)***
RETA –.3395 .0675 –.5313 .0038
(.28) (.65)
OINS .163 .1874 .1077 .0129
(.52) (.39)
WCTA –.1228 .2046 –.1171 .1985
(7.12)*** (6.8)***
–2 Log L 259.070 253.330
Model Chi-Square 60.06*** 65.80***
Note: For model 1 , where . Forprob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
Z a bR c Xi i j j i
j
= + +∑$
,
model 2 , where .prob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
Z a b c X ei k k i j j i i
jk
= + + +∑∑ β , ,
is the estimated monthly stock return of the firm i. PMI represents the systematic risks or
$
Ri
the sensitivity of a firm to the changes in industrial production . CPI represents the systematic
risks or the sensitivity of a firm to the changes in consumer price index. INT represents the
systematic risks or the sensitivity of a firm to the changes in interest rates. MS2 represents the
systematic risks or the sensitivity of a firm to the changes in M2 money supply. SETA
represents the ratio of stockholders’ equity to total assets. RETA represents the ratio of
retained earnings to total assets. OINS represents the ratio of operating income to net sales.
WCTA represents the ratio of working capital to total assets. Parentheses include the standard
errors in [ ], t-values, for H0: variances are equal, in ( ), and F-values, for H0: means are not
significantly different from zero, in { }. ***, **, and * denote statistical significance at the
Investigation of Financial Distress for Thai Firms 119
TABLE 4. The Full-sample Test for Model 1
Distress Prediction Non-distressed Prediction Total Prediction
Accuracy Accuracy Accuracy
Cutoff
Point Amount % Amount % Amount %
0 55 100.00% 0 0.00% 55 13.89%
.1 48 87.27% 205 60.12% 253 63.89%
.2 33 60.00% 294 86.22% 327 82.58%
.3 21 38.18% 323 97.95% 344 86.87%
.4 13 23.64% 334 97.95% 347 87.63%
.5 8 14.55% 336 98.53% 344 86.87%
.6 6 10.91% 338 99.12% 344 86.87%
.7 4 7.27% 339 99.41% 343 86.62%
.8 3 5.45% 339 99.41% 342 86.36%
.9 2 3.64% 339 99.41% 341 86.11%
1 0 0.00% 341 100.00% 341 86.11%
Note: For model 1 , where ,prob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
,
ˆ
i i j j iZ a bR c X= + ∑
. The sample consists of 396 firms in which 341 firms are non-0, ,
ˆ ˆ
i i k iR Fβ β= + ∑
distressed and 55 firms are financially distressed in 1997.
TABLE 5. The In-sample Test for Model 1
Distress Prediction Non-distressed Prediction Total Prediction
Accuracy Accuracy Accuracy
Cutoff
Point Amount % Amount % Amount %
0 30 100.00% 0 0.00% 30 12.99%
.1 25 83.33% 154 76.62% 179 77.49%
.2 15 50.00% 185 92.04% 200 86.58%
.3 15 50.00% 195 97.02% 210 90.91%
.4 12 40.00% 199 99.01% 211 91.34%
.5 10 33.33% 199 99.01% 209 90.48%
.6 10 33.33% 199 99.01% 209 90.48%
.7 6 20.00% 199 99.01% 205 88.75%
.8 5 16.67% 199 99.01% 204 88.31%
.9 4 13.33% 199 99.01% 203 87.88%
1 0 0.00% 201 100.00% 201 87.01%
Note: For model 1 , where ,prob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
,
ˆ
i i j j iZ a bR c X= + ∑
. The sample is divided into two subsamples: 60% of the full sample as0, ,
ˆ ˆ
i i k iR Fβ β= + ∑
the in-sample (learning sample) and 40% of the full sample as the out-sample (holdout sample).
Multinational Finance Journal120
TABLE 6. The Out-sample Test for Model 1
Distress Prediction Non-distressed Prediction Total Prediction
Accuracy Accuracy Accuracy
Cutoff
Point Amount % Amount % Amount %
0 25 100% 0 0.0% 25 15.15%
.1 22 88% 103 73.57% 125 75.76%
.2 18 72% 120 85.71% 138 83.64%
.3 13 52% 126 90.00% 139 84.24%
.4 11 44% 133 95.00% 144 87.27%
.5 11 44% 135 96.43% 146 88.49%
.6 5 20% 136 97.14% 141 85.46%
.7 3 12% 137 97.86% 140 84.85%
.8 2 8% 137 97.86% 139 84.24%
.9 2 8% 137 97.86% 139 84.24%
1 0 0% 140 100.00% 140 84.85%
Note: For model 1 , where ,prob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
,
ˆ
i i j j i
Z a bR c X= + ∑
. The sample is divided into two subsamples: 60% of the full sample as0 , ,
ˆ ˆ
i i k i
R Fβ β= + ∑
the in-sample (learning sample) and 40% of the full sample as the out-sample (holdout sample).
TABLE 7. The Full-sample Test for Model 2
Distress Prediction Non-distressed Prediction Total Prediction
Accuracy Accuracy Accuracy
Cutoff
Point Amount % Amount % Amount %
0 25 100% 0 0.00% 25 15.15%
.1 22 88% 103 73.57% 125 75.76%
.2 18 72% 120 85.71% 138 83.64%
.3 13 52% 126 90.00% 139 84.24%
.4 11 44% 133 95.00% 144 87.27%
.5 11 44% 135 96.43% 146 88.49%
.6 5 20% 136 97.14% 141 85.46%
.7 3 12% 137 97.86% 140 84.85%
.8 2 8% 137 97.86% 139 84.24%
.9 2 8% 137 97.86% 139 84.24%
1 0 0% 140 100.00% 140 84.85%
Note: For model 2 , where,prob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
, , .i k k i j j i i
k j
Z a b c X eβ= + + +∑ ∑
The sample consists of 396 firms in which 341 firms are non-distressed and 55 firms are financially
distressed in 1997.
Investigation of Financial Distress for Thai Firms 121
TABLE 9. The Out-sample Test for Model 2
Distress Prediction Non-distressed Prediction Total Prediction
Accuracy Accuracy Accuracy
Cutoff
Point Amount % Amount % Amount %
0 30 100.00% 0 0.00% 30 12.99%
.1 26 86.67% 153 76.12% 179 77.49%
.2 17 56.67% 183 91.05% 200 86.58%
.3 15 50.00% 195 97.02% 210 90.91%
.4 13 43.33% 199 99.01% 212 91.78%
.5 11 36.67% 199 99.01% 210 90.91%
.6 9 30.00% 199 99.01% 208 90.04%
.7 8 26.67% 199 99.01% 207 89.61%
.8 5 16.67% 199 99.01% 204 88.31%
.9 4 13.33% 200 99.50% 204 88.31%
1 0 0.00% 201 100.00% 201 87.01%
Note: For model 2 , whereprob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
, , .i k k i j j i i
k j
Z a b c X eβ= + + +∑ ∑
The sample is divided into two subsamples: 60% of the full sample as the in-sample (learning
sample) and 40% of the full sample as the out-sample (holdout sample).
TABLE 8. The In-sample Test for Model 2
Distress Prediction Non-distressed Prediction Total Prediction
Accuracy Accuracy Accuracy
Cutoff
Point Amount % Amount % Amount %
0 30 100.00% 0 0.00% 30 12.99%
.1 26 86.67% 153 76.12% 179 77.49%
.2 17 56.67% 183 91.05% 200 86.58%
.3 15 50.00% 195 97.02% 210 90.91%
.4 13 43.33% 199 99.01% 212 91.78%
.5 11 36.67% 199 99.01% 210 90.91%
.6 9 30.00% 199 99.01% 208 90.04%
.7 8 26.67% 199 99.01% 207 89.61%
.8 5 16.67% 199 99.01% 204 88.31%
.9 4 13.33% 200 99.50% 204 88.31%
1 0 0.00% 201 100.00% 201 87.01%
Note: For model 2 , whereprob Y
Z
i
i
= =
+ −
1
1
1
0 5 0 5exp
, , .i k k i j j i i
k j
Z a b c X eβ= + + +∑ ∑
The sample is divided into two subsamples: 60% of the full sample as the in-sample (learning
sample) and 40% of the full sample as the out-sample (holdout sample).
Multinational Finance Journal122
12. Unlike the logit coefficients, the elasticity is independent of measurement units for
the variables.
13. Type I errors is the probability of a failed firm misclassified as an unfailed firm, and
type II error is the probability of an unfailed firm misclassified as a failed firm. can be
defined as the misclassification of the failed firm into the unfailed firm and type II errors can
be defined as the misclassification of the unfailed firm into the failed firm.
lower the probability of financial distress. Also, table 3 reports the
elasticities, as defined in Theodossiou, et al. (1996), which measure the
importance of the variables in the model.12
It can be seen that SETA
and , for model 1, and SETA and bCPI , for model 2, are the most$Ri
important variables, consistent with the Wald chi-square statistics.
Tables 4-9 show the predictive ability of the macro-related micro-
crisis prediction models at various cut-off points. Since type I errors are
more costly than type II errors13
, the study attempts to determine a cut-
off point that correctly discriminates the distressed firms at a greater
number than does the non-distressed firms or, in other words, in which
type I errors are approximately smaller than type II errors. On average,
the cut-off point at .1 seems to be optimal, indicating that a firm that has
a probability of financial distress of 10% or over will be classified as a
distressed firm and a firm that has a probability under 10% will be
classified as a non-distressed firm. The prediction results for the
indirect model (equation 1) are shown in table 4-6. At the cut-off point
of .1, the total prediction accuracy equals to 63.89% for the full-sample
test, 77.49% for the in-sample test, and 75.76% for the out-sample test
as shown in table 4-6. The prediction results for the direct model
(equation 2) are shown in table 7-9. At the cut-off point of .1, the total
prediction accuracy equals to 68.43% for the full-sample test, 77.49%
for the in-sample test, and 70.91% for the out-sample test as shown in
table 7-9. The out-sample predictive ability of over 70% for both
models signifies promising outcomes of this study.
V. Conclusions and Discussion
This article employed the 1997 economic crisis in Thailand as a case
study. Using logit analysis, the model attempts to portray corporate
bankruptcy caused by economic collapse as systematic risk. The macro-
related micro-crisis exploration model developed in this study has two
specifications. The first specification (the indirect test) uses firms’
Investigation of Financial Distress for Thai Firms 123
estimated stock returns calculated from the estimated changes in
macroeconomic variables and firms’ sensitivities to those macro
variables as proxies of macro-related factors, while the second
specification (the direct test) uses firms’ sensitivities to macro variables
as proxies for macroeconomic factors. The contribution of the model is
to bridge a firm’s sensitivity to macroeconomic conditions and its
financial characteristics in order to investigate the firm’s financial
distress.
The results are promising and confirm that macroeconomic
conditions are important factors in determining a firm’s possibility of
financial crises. The estimated stock return variable shows significance
in the indirect-test specification (equation 1) and the sensitivity-to-
inflation variable is also significant at the .01 level in the direct-test
specification (equation 2). Since 1996-1997, the Thai economy showed
signs of a steep decline causing negative returns for equity investors.
The empirical findings demonstrate that, in an economic crisis, the
higher or less negative a firm’s return, the lower the probability of the
firm’s financial distress. Inflation is the most crucial variable because
the higher a firm’s sensitivity to inflation, the more likely the firm’s
exposure to financial distress.
For testing the predictive ability of the model, the study uses the cut-
off point at .1 to separate financially distressed firms from non-
distressed firms because type I errors are quite costly, relative to type II
errors. For both specifications, the predictive ability or accuracy ranges
between 60-80% of the sample. An out-sample test of predictive ability
confirms that the model has an accuracy of over 70% of the holdout
sample.
In conclusion, the article develops the investigation model using the
1997 Thailand financial crisis. The model incoporates the firms’
sensitivitytomacroeconomicvariablesandtheirfinancialcharacteristic.
It will be useful in providing warning signals of upcoming crises so that
policy makers can seek protective measure in order to immunize the
economy and protect against contagious, potentially lethal financial
diseases.
References
Alba, P.; Claessens, S.; and Djankov, S. 1998. Thailand’s Corporate financing
and governance structures: Impact on firms’ competitiveness. Conference
Multinational Finance Journal124
onthailand’sdynamiceconomicrecoveryandcompetitiveness.May20-21,
1998, UNCC, Bangkok, Thailand.
Altman, E. I. 1968. Financial ratios, discriminant analysis and the prediction of
corporatebankruptcy.TheJournalofFinance23(September):589-609.
Altman, E. I. 1983. Multidimensional graphics and bankruptcy prediction: a
comment. Journal of Accounting Research 21(Spring): 297-299.
Altman, E. I. 1984. The Success of business failure prediction models: an
international survey. Journal of Banking and Finance 8: 171-198.
Altman, E. I. 1994. Corporate distress diagnosis: Comparisons using linear
discriminant analysis and neural network (the Italian experience). Journal
of Banking and Finance 18: 505-529.
Altman, E. I. and Saunders, A. 1998. Credit risk measurement: Developments
over the last 20 years. Journal of Banking and Finance 21: 1721-1742.
Barnes, P. 1987. The analysis and use of financial ratios: A review article.
Journal of Business Finance and Accounting 14: 449-461.
Beaver, W. H. 1966. Financial ratios as predictors of failure. Empirical
research in accounting: selected studies. Journal of Accounting Research
4 (Supplement).
Caprio, G. and Klingebiel, D. 1996. Dealing with bank insolvencies: Cross
countryexperience, The World Bank, Washington, D.C.
Corsetti, G.; Pesenti, P.; and Roubini N. 1998. What Caused the Asian
currency and financial crisis? Unpublished Manuscript.
Demirguc-Kunt, A. and Detragiache, E. 1997. The determinants of banking
crises: Evidence from developing and developed countries. International
Monetary Fund Working Paper.
Eichengreen, B.; Rose, A. K.; and Wyplosz, C. 1996. Contagious currency
crises. NBER Working Paper No. 5681.
Eichengreen, B. and Rose, A. K. 1998. Staying Afloat when the wind shifts:
externalfactorsandemerging-marketbankingcrises.NBERWorkingPaper
No.6370.
Flagg, J. C.; Giroux, G. A.; and Wiggins, C. E. Jr. 1991. Predicting corporate
bankruptcy using failing firms. Review of Financial Economics 1: 67-78.
Flood, R. and Marion, N. 1998. Perspectives on the Recent Currency Crisis
Literature. NBER Working Paper No. 6380.
Gilbert, L. R.; Krishnagopal, M.; and Schwartz, K. B. 1990. Predicting
bankruptcy for firms in financial distress. Journal of Business Finance and
Accounting 17 (Spring):161-171.
Glick, R. and Rose, A. K. 1998. Contagion and trade: Why are currency crises
regional? NBER Working Paper No. 6806.
Hopwood, W.; McKeown , J. C. ; and Mutchler, J. F. 1989. A Test of the
incremental explanatory power of opinions qualified for consistency and
uncertainty. The Accounting Review 66:28-48.
Johnsen, T. and Melicher, R. W. 1994. Predicting Corporate bankruptcy and
financial distress: information value added by multinomial ligit models.
Journal of Economics & Business 46:269-286.
Investigation of Financial Distress for Thai Firms 125
Kaminsky, G. and Reinhart, C. M. 1996. The Twin crises: The causes of
banking and balance-of-payment problems. Federal reserve board and
international monetary fund. Unpublished Manuscript.
Kaminsky, G.; Lizondo, S.; and Reinhart, C. M. 1997. Leading indicators of
currency crises. International monetary fund. Working Paper.
La Porta, R.; Lopez-de-Silanes, F. ; Shleifer, A.; and Vishny, R. W. 1998. Law
and finance. Journal of Political Economy, forthcoming.
Maddala, G. S. 1986. Econometric issues in the empirical analysis of thrift
institutions’ insolvency and failure. Invited Research Working Paper 56,
Federal Home Loan Bank Board, October.
Mensah, Y. M. 1983. The differential bankruptcypredictive abilityof specific price
level adjustments: Some empirical evidence. The Accounting Review
LVIII(April):228-246.
Ohlson, J. A. 1980. Financial ratios and the probabilistic prediction of
bankruptcy. Journal of Accounting Research 18(Spring):109-130.
Platt, H. E. and Platt, M. B. 1990. Improving ex ante bankruptcy classification
using stable predictive variables. Journal of Business Finance and
Accounting 17:31-51.
Platt, H. E. and Platt, M. B. 1991. A note on the use of industry-relative ratios
in bankruptcy prediction. Journal of Banking and Finance 15:1183-1194.
Poston, K. M.; Harmon, W. K.; and Gramlich, J. D. 1994. A test of financial
ratios as predictors of turnaround versus failure among finanially
distressed firms. Journal of Applied Business Research 10: 41-56.
Rachel, S. and Sachs, J. 1998. The onset of the east Asian financial crisis.
NBER Working Paper No. 6680.
Rose, R. S.; Andrews, W. T.; and Giroux, G. A. 1982. Predicting business
failure: A macroeconomic perspective. Journal of Accounting, Auditing,
and Finance (Fall): 20-31.
Salchenberger, L. M.; Cinar, E. M.; and Lash, N. A. 1992. Neural networks:
A new tool for predicting thrift failures. Decision Sciences 23: 899-916.
Scott, J. 1981. The probability of bankruptcy: A comparison of empirical
prediction and theoretical model. Journal of Banking and Finance 5: 317-
344.
Shivaswamy, M.; Hoban, J. P.; and Matsumoto, K. 1993. A behavioral analysis
of financial ratios. The Mid-Atlantic Journal of Business 29 (March) :7-24.
Theodossiou, P.; Kahya, E. ; Saidi, R.; and Philippatos, G. 1996. Financial
distress and corporate acquisitions: Further empirical evidence. Journal of
Business Finance and Accounting 23 (July): 699-719.
Thomson, J. B. 1992. Modelingthe bank regulator’s closure option: A two-step
logit regression approach. Journal of Financial Services Research.
Zavgren, C. 1983. The prediction of corporate failure: The state of the art.
Journal of Accounting Literature 2: 1-35.
Zmijewski, M. E. 1984. Methodological issues related to the estimation of
financial distress prediction models. Journal of Accounting Research 22
(Supplement): 59-85.

More Related Content

PPTX
Dba bvasconez dherrera Cluster
PDF
11.relationships between indian and other south east stock markets
PDF
Factors affecting stock market prices in amman stock exchange
PDF
11.how can behavioural finance help us in better understanding the recent glo...
DOCX
Evaluation of the Development and Performance of Selected GCC and Non-GCC St...
PDF
SME development, constraints, credit risk & islamic banking solutions
PPTX
Mishkin fmi9ge ppt_c02
PDF
The Investment Climate for Japanese Manufacturing MNCs in Asia: ICSEAD's In...
Dba bvasconez dherrera Cluster
11.relationships between indian and other south east stock markets
Factors affecting stock market prices in amman stock exchange
11.how can behavioural finance help us in better understanding the recent glo...
Evaluation of the Development and Performance of Selected GCC and Non-GCC St...
SME development, constraints, credit risk & islamic banking solutions
Mishkin fmi9ge ppt_c02
The Investment Climate for Japanese Manufacturing MNCs in Asia: ICSEAD's In...

What's hot (20)

DOCX
DETERMINANTS OF THE CAPITAL STRUCTURE OF THE COMPANIES OF THE FOOD AND BEVERA...
PDF
11.relationships between indian and other south east stock markets
PDF
Relationships between indian and other south east stock markets
PDF
Tangible market information and stock returns the nepalese evidence synopsis
PDF
Case private equity
PDF
Short Selling: Cleaning Up After Elephants
PDF
Stock market anomalies a study of seasonal effects on average returns of nair...
DOCX
Analysis of Stock Market Anomalies worldwide
PDF
Market efficiency, market anomalies, causes, evidences, and some behavioral a...
PDF
Emerging Stock Markets and Performance of IPOs: An Application to the Regiona...
PPTX
How stock market impact on economy
PDF
ABFER Research Digest 2013
PDF
CASE Network Studies and Analyses 211 - Theoretical Aspects of Currency Crises
PDF
PDF
Determinants of capital_structure_an_emp
PPTX
Mishkin fmi9ge ppt_c01
DOCX
19 Ratio Analysis
PDF
SME Manufacturing Credit Risk Model Forecast Correctness and Result of Model
PDF
From Export Specialization in Natural resources to Diversification in Manufac...
DETERMINANTS OF THE CAPITAL STRUCTURE OF THE COMPANIES OF THE FOOD AND BEVERA...
11.relationships between indian and other south east stock markets
Relationships between indian and other south east stock markets
Tangible market information and stock returns the nepalese evidence synopsis
Case private equity
Short Selling: Cleaning Up After Elephants
Stock market anomalies a study of seasonal effects on average returns of nair...
Analysis of Stock Market Anomalies worldwide
Market efficiency, market anomalies, causes, evidences, and some behavioral a...
Emerging Stock Markets and Performance of IPOs: An Application to the Regiona...
How stock market impact on economy
ABFER Research Digest 2013
CASE Network Studies and Analyses 211 - Theoretical Aspects of Currency Crises
Determinants of capital_structure_an_emp
Mishkin fmi9ge ppt_c01
19 Ratio Analysis
SME Manufacturing Credit Risk Model Forecast Correctness and Result of Model
From Export Specialization in Natural resources to Diversification in Manufac...
Ad

Similar to 01 journal-financial distress using macro and micro (20)

PDF
Distress risk and stock returns in an emerging market
PDF
Investigation Of Academic Thesis And Articles Produced In The Field Of Financ...
PDF
TCQT_NHOM_8__SLIDE_THUYET_TRINH.pdf hkknbjp
PDF
Klibel5 bus a1
PPTX
Meeting 9 - Bankruptcy (Financial Management)
PDF
F3123642
PPTX
Factors that affect financial distress
PDF
35-501-1-SP The Zmijewski analysis i.pdf
PDF
An in depth analysis of the altman’s failure prediction model
PDF
Bankruptcy prediction final report by sanapathi kiran kumar
PPTX
Suchorita case study eco_mba_feb 2021
PDF
Predicting Corporate Failure - An Application of Discriminate Analysis
PPTX
Presentation for Gracia Caroline Finishing
PDF
Establishing the-effectiveness-of-market-ratios-in-predicting-financial-distr...
PDF
Establishing the effectiveness of market ratios in predicting financial distr...
PDF
Q35122127
PPTX
Econs powerpoint 2
PDF
Measuring
PDF
Studio sulla capacità del modello predittivo del fallimenti Altman Z-Score ne...
PDF
Financial Distress Corporate Restructuring and Firm Survival An Empirical Ana...
Distress risk and stock returns in an emerging market
Investigation Of Academic Thesis And Articles Produced In The Field Of Financ...
TCQT_NHOM_8__SLIDE_THUYET_TRINH.pdf hkknbjp
Klibel5 bus a1
Meeting 9 - Bankruptcy (Financial Management)
F3123642
Factors that affect financial distress
35-501-1-SP The Zmijewski analysis i.pdf
An in depth analysis of the altman’s failure prediction model
Bankruptcy prediction final report by sanapathi kiran kumar
Suchorita case study eco_mba_feb 2021
Predicting Corporate Failure - An Application of Discriminate Analysis
Presentation for Gracia Caroline Finishing
Establishing the-effectiveness-of-market-ratios-in-predicting-financial-distr...
Establishing the effectiveness of market ratios in predicting financial distr...
Q35122127
Econs powerpoint 2
Measuring
Studio sulla capacità del modello predittivo del fallimenti Altman Z-Score ne...
Financial Distress Corporate Restructuring and Firm Survival An Empirical Ana...
Ad

Recently uploaded (20)

PDF
Buy Verified Payoneer Accounts for Sale - Secure and.pdf
PDF
CLIMATE CHANGE AS A THREAT MULTIPLIER: ASSESSING ITS IMPACT ON RESOURCE SCARC...
PDF
5-principles-of-PD-design.pdfvvvhvjvvcjcxhhcjb ggfvjhvjjhbvbbbvccxhgcxzzghjbv...
PDF
HCWM AND HAI FOR BHCM STUDENTS(1).Pdf and ptts
PDF
2018_Simulating Hedge Fund Strategies Generalising Fund Performance Presentat...
PDF
Fintech Regulatory Sandbox: Lessons Learned and Future Prospects
PPTX
Module5_Session1 (mlzrkfbbbbbbbbbbbz1).pptx
PDF
Buy Verified Stripe Accounts for Sale - Secure and.pdf
PPTX
ML Credit Scoring of Thin-File Borrowers
PDF
Pitch Deck.pdf .pdf all about finance in
PPTX
2. RBI.pptx202029291023i38039013i92292992
PPT
CompanionAsset_9780128146378_Chapter04.ppt
PDF
Truxton Capital: Middle Market Quarterly Review - August 2025
PPTX
Group Presentation Development Econ and Envi..pptx
PPTX
Machine Learning (ML) is a branch of Artificial Intelligence (AI)
PDF
Pension Trustee Training (1).pdf From Salih Shah
PPTX
Very useful ppt for your banking assignments Banking.pptx
PDF
Lundin Gold Corporate Presentation August 2025
PPTX
OAT_ORI_Fed Independence_August 2025.pptx
PPT
KPMG FA Benefits Report_FINAL_Jan 27_2010.ppt
Buy Verified Payoneer Accounts for Sale - Secure and.pdf
CLIMATE CHANGE AS A THREAT MULTIPLIER: ASSESSING ITS IMPACT ON RESOURCE SCARC...
5-principles-of-PD-design.pdfvvvhvjvvcjcxhhcjb ggfvjhvjjhbvbbbvccxhgcxzzghjbv...
HCWM AND HAI FOR BHCM STUDENTS(1).Pdf and ptts
2018_Simulating Hedge Fund Strategies Generalising Fund Performance Presentat...
Fintech Regulatory Sandbox: Lessons Learned and Future Prospects
Module5_Session1 (mlzrkfbbbbbbbbbbbz1).pptx
Buy Verified Stripe Accounts for Sale - Secure and.pdf
ML Credit Scoring of Thin-File Borrowers
Pitch Deck.pdf .pdf all about finance in
2. RBI.pptx202029291023i38039013i92292992
CompanionAsset_9780128146378_Chapter04.ppt
Truxton Capital: Middle Market Quarterly Review - August 2025
Group Presentation Development Econ and Envi..pptx
Machine Learning (ML) is a branch of Artificial Intelligence (AI)
Pension Trustee Training (1).pdf From Salih Shah
Very useful ppt for your banking assignments Banking.pptx
Lundin Gold Corporate Presentation August 2025
OAT_ORI_Fed Independence_August 2025.pptx
KPMG FA Benefits Report_FINAL_Jan 27_2010.ppt

01 journal-financial distress using macro and micro

  • 1. *Financial support for this research was provided by the Listed Company Association, Thailand. The authors appreciate comments and suggestions by four anonymous referees and participants at presentationsat theThird AnnualEuropean FMAConference (Barcelona 1999), the Sixth Annual Conference of the Multinational Finance Society (Toronto 1999), and the Eleventh Annual PACAP/FMA Finance Conference (Singapore 1999). (Multinational Finance Journal, 1999, vol. 3, no. 2, pp. 103–125) © Multinational Finance Society, a nonprofit corporation. All rights reserved. 1 An Investigation of Thai Listed Firms’ Financial Distress Using Macro and Micro Variables* Sunti Tirapat Chulalongkorn University, Thailand Aekkachai Nittayagasetwat National Institute of Development Administration, Thailand The emergence of the economic crisis in Thailand in 1997 is an interesting case for academic studies. Internationally, it had a contagion effect, spreading to countries in Asia and in other regions. Domestically, it caused a great many industrial and corporate bankruptcies. The Thai economy had been relatively stable since 1984. The recent development in 1997, however, produced a sudden economic slumpresultinginclosuresof manyThai corporations. Using alogitregression,thisstudydevelopsamacro-relatedmicro-crisisinvestigation model. The significance of the model is in its ability to bridge a firm’s sensitivity to macroeconomic conditions and its financial characteristics in order to explore a firm’s financial distress. The findings indicate that macroeconomicconditionsarecriticalindicatorsof potentialfinancialcrisisfor a firm. The article shows that the higher a firm’s sensitivity to inflation, the higher the firm’s exposure to financial distress. Keywords: bankruptcy, financial distress, prediction model, and Thailand crisis. I. Introduction In 1997, an Asian financial crisis emerged. From Thailand, the crisis rapidly spread throughout the region. The swiftness and the confounding effects of the “Tom Yam Kung” crisis have raised several
  • 2. Multinational Finance Journal104 1. “Tom Yam Kung” is a famous Thai dish and it became the name of the 1997 economic crisis spreading from Thailand. 2. The speculative attack literature is reviewed by Flood and Marion (1998). The contagious nature of the crisis has been investigated by Eichengreen, Rose, and Wyplosz (1996) and more recently by Glick and Rose (1998). 3. During 1992-1995, the growth of average net incomes of firms listed in the Stock Exchange of Thailand was around 22% while in the 1996 it declined to – 31.3%. 4. Studies that use only financial ratios include early works such as Beaver (1966) and Altman (1968). Much of the recent work is discussed in Altman and Saunders (1998). Studies that incorporate accountant opinions include Hopwood et al. (1989) and Flagg et al. (1991). Studies that use only financial ratios include early works such as Beaver (1966) and Altman (1968), and Altman, Haldeman, and Narayanan (1977). Much of the recent work is discussed in Altman and Saunders (1998) and Zavgren (1983). Studies that incorporate accountant opinions include Hopwood et al. (1989), Flagg et al. (1991). 5. For example, in Thailand fewer than 10 companies listed in the Stock Exchange of Thailand were financially distressed during 1987-1994. interesting economic questions.1 What was the root of the problem? Why was the crisis ‘contagious’? What triggered the speculative attack? These issues have already been investigated by prominent economic scholars. Numerous causes of the crisis have been proposed, ranging from the role of financial panic (Rachel and Sachs [1998]) to a nexus of moral hazard problems (Corsetti, Pensenti, and Roubini [1998]).2 The crisis, however, has provided an opportunity to investigate another area of interest, the role of corporate bankruptcy. The impact of the crisis has resulted in a large number of Thai firms going bankrupt in 1997.3 Since the massive bankruptcy is an exceptional case in Thailand, this article aims to seek the cause of this event. Prior articles on predicting bankruptcy focus primarily on financial variables such as financial ratios, cash flows, or accountingopinions.4 Many studies have been conducted in developed countries because data are readily available. In emerging markets, however, this line of research is scarce since, in growing economies, bankrupt firms are actually rare.5 The crisis provides us with an opportunity to investigate the prediction model in developing countries such as Thailand. An investigation of the financial distress model may be of interest and provide some insights into this area of research since the Thai environment may differ from that of the developed markets on several aspects. First, the institutional structure of the financial sector in Thailand can be characterized as a bank-centered financial system, in contrast with that of a market-based
  • 3. Investigation of Financial Distress for Thai Firms 105 6. See La porta, Lopez-de-Silanes, Shleifer, and Vishy (1998), which compile ownership concentration of 45 developed and developing countries. 7. Alba,Claessens,andDjankov(1998)pointoutthataccountingandauditingpractices in Thailand may be due to the lack of strong self-regulatory organizations in the auditing and accounting profession and a shortage of well-qualified accountants and auditors. system such as that in the U.S. The credits from financial institutions (commercial banks and finance companies) are about 5.5 times larger than funds raised (in form of equity and bonds) from capital markets in 1994 and 16.7 times larger than funds raised in 1997. Secondly, the ownership structure in Thailand is highly concentrated. The average percentage of common shares owned by the three largest shareholders in the ten largest firms is around 44 percent.6 The typical Thai conglomerate is controlled by family and highly leveraged by local financial institutions. This structure has led to inefficient investments before the 1997 financial crisis. The ratio of earnings before interest and taxes (EBIT) over interest expenses declined from 5.78 in 1994 to 1.49 in 1997, while the leverage ratio rose from 1.5 to 2.95 during the same period. Thirdly, anecdotal evidence shows that although the accounting standards in Thailand are similar to those of other international standards such as the International Accounting Standard (IAS) or the Financial AccountingStandardBoard(FASB),theaccountingpractices are of concern.7 Hence, it is interesting to see whether or not financial ratios can appropriately explain the causes of failure of firms in developing countries such as Thailand. In addition, an investigation of financially distressed firms in emerging markets, in our view, should take economic conditions into account since these firms are typically small and operating in an open economy. Moreover, as a firm’s accounting system is subject to accounting standard, e.g., Generally Accepted Accounting Principles (GAAP), it seldom reflects immediate changes in a firm’s financial situation when a crisis strikes the overall economy. Therefore, accountingvariablesalonemaynotadequatelyexplainafirm’sfinancial distress, and the firm’s sensitivity to macroeconomic factors should be taken as a significant explanatory variable to indicate a firm’s financial distress. In other words, firms that are more sensitive to economic shocks should be more prone to financial distress than firms that are less sensitive. This article attempts to incorporate the macroeconomic factors in the investigation of financially distressed firms. The purpose of incorporating macroeconomic variables is to capture the sensitivity
  • 4. Multinational Finance Journal106 to change in macroeconomic factors. This perspective, in our view, is more consistent with the Thai situation. Using the two-stop logit approach,thisstudydevelops a macro-relatedmicro-crisisinvestigation model. It is found that the sensitivity of firms to economic conditions playsamajorroleindifferentiatingthe financiallydistressedcompanies from the non-distressed ones. The article is organized as follows: Section IIreviews prior research. The sample and methodology are discussed in section III. Empirical results of the model are reported in section IV. The discussion of the findings and conclusions are presented in section V. II. Literature Review A. Prediction of Bankruptcy Bankruptcy prediction models were pioneered by Beaver’s (1966) univariate test and Altman’s (1968) multivariate discriminant analysis. Both articles document that financial variables can be used to predict bankruptcy. Since then, prediction of corporate failure has been a topic of much interest and more recent works have extended this line of research in three areas: statistical techniques, definitions of bankruptcy, and a greater variety of explanatory variables. For example, Ohlson (1980) utilized the logit and probit analysis to estimate the probability of bankruptcy. The adjustments of estimation bias resulting from oversampling, such as weights based on prior probabilities in the estimationprocessandoptimalcut-offpoint,arediscussedinZmijewski (1984). The problems of pooling data due to a small number of samples are addressed in Zavgren (1983) . The second area of extension has dealt with the definitions of bankruptcy. For example, a model that distinguishes between financially distressed firms that survive and financially distressed firms that ultimately go bankrupt has been investigated. Gilbert, Menon, and Schwartz (1990) find different explanatory financial variables for these two groups of firms. In addition, a sample firm may be classified into more than two categories (bankrupt or nonbankrupt) and the classification probabilities can be estimated by the multinomial logit technique. For example, Poston, Harmon, and Gramlich (1994) assigned firms into one of three groups according to each firm's financial condition: turnarounds, business failures, and survivors. They
  • 5. Investigation of Financial Distress for Thai Firms 107 8. The specification of their models includes four financial ratios, sales growth, and the interaction of industry output and financial ratios. 9. Caprio and Klingebiel (1996) document 49 banking crises in the 1970s and 1980s; and 33 crises occurred over the first six years of the present decade. find that financial ratios are not so useful in distinguishing between financially distressed firms that are able to turn around and those that are unable to avoid failure. Johnsen and Melicher (1994), on the other hand, suggest that by using multinomial logit models, the classification errors can be significantly reduced. The third area involves some adjustment of explanatory variables either by covering additional variables other than financial ratios or industry-adjusted ratios. For example, Hopwood, McKeown, and Mutchler (1989) and Flagg, Giroux, and Wiggins (1991) find that a qualified opinion is significant in distinguishing financially distressed firms. Some studies include macroeconomic variables to control for changes in the business environment. Rose and Giroux (1982) examine 28 business cycle indicators and find that economic conditions affect the failure process. Mensah (1983) evaluates the bankruptcy model usingtheprice-leveladjusted (SPL) data. ThefindingsindicatethatSPL data do not significantly improve bankruptcy prediction. Finally, Platt and Platt (1990) control for industry differences by using industry- normalizing ratios. The industry-relative framework results in stable bankruptcymodels. Therefore,industrialgrowthhasasignificanteffect on corporate failure. Furthermore, Platt and Platt (1991) investigate the stability and completeness of a bankruptcy model based on industry- relative ratios compared to that of unadjusted ratios.8 Both models yield stable parameter estimation. The industry- relative specification, however, appears to provide incremental information not contained in the model based on the unadjusted ratios. B. Prediction of Currency and Banking Crises Another strand of related research concerns the prediction of currency and banking crises; for example, Demirguc-Kunt and Detragiache (1997), Eichengreen and Rose (1998). These studies are motivated by the recent spate of crises9 in Europe (Finland, Sweden), Latin America (Mexico, Argentina) and Asia (Thailand, Indonesia). Kaminsky and Reinhart (1996), in analyzing a sample of 20 crises, report the importance of macroeconomic factors in a crisis. They document that
  • 6. Multinational Finance Journal108 output, the stock market, and the real exchange rate usually peak about a year before the onset of a banking crisis. Real interest rates and bank deposits rise in the period leading up to the crisis. Demirguc-Kunt and Detragiache (1997) applya multivariate logit model to the determinants of banking in a sample of developed as well as developing countries between 1980 and 1994. They find that in general, low growth and high inflation are associated with the probability of a banking crisis. EichengreenandRose(1998)restricttheir sample to cover onlybanking crises in emerging markets from 1975 through 1992. They apply a multivariate probit model to nine economic variables; four of their variables are “international” in nature,three are “domestic”, and two are “external” in nature. Their results suggest that banking crises in emerging countries tend to occur in response to a conjuncture of unfavorable developments in international markets and domestic imbalances. Specifically, rising industrial-country interest rates precipitate banking crises: a one-percent increase in interest rates is associated with an increase in the probability of banking crises of around three percent. Kaminsky, Lizondo, and Reinhart (1997) examine the empirical evidence on 25 currency crises and compare the merits of alternative approaches in providing early indications of the crises. They document that the best track record includes exports, deviations of the real exchange rate from the trend, the ratio of broad money supply to gross international reserves, output, and equity prices. Kaminsky and Reinhart (1996) also investigate 76 currency crises and 26 banking crises and find that banking and currency crises are closely related in the aftermath of financial liberalization, with banking crises generally preceding currency crashes. III. Data and Methodology A. Sample and Data The sample includes firms listed in the Stock Exchange of Thailand (SET)thatexperiencedfinancialdistressin1997. Financiallydistressed firms are defined in this article as firms that either (1) were closed down by governmental authorities (all of which were banks and finance companies) or (2) were required by the Bank of Thailand or SET to submit restructuring plans (these companies were designated as C or
  • 7. Investigation of Financial Distress for Thai Firms 109 10. The SET posts a supervision sign of “C” (the filing for compliance) or “SP*” (the suspensionor thetemporaryprohibitionof tradinguntil thecauses of delisting areeliminated) against a listed company and the firm must submit the SET documents supporting the company’s financial position. SP10 by the Exchange). There were 459 firms listed in the SET in 1996, of which 55 were financially distressed by this definition. Firms that were first listed on the SET in 1995 or later and firms with incomplete data are excluded. The resulting sample consists of 341 non-financially distressed firms and 55 financially distressed firms. The characteristics of firms in the sample are reported in table 1. B. Method Although some prior research, e.g., Rose, Andrews, and Giroux (1982), have included macroeconomic variables in the prediction model, this study differs on one important aspect. The purpose of incorporating macroeconomic variables is not to control for changes in the business environment but rather to capture the sensitivity to change in macroeconomic factors. This perspective, in our view, is more consistent with the Thai situation. The sensitivity of firms to economic conditions is expected to play a major role in differentiatingthe financiallydistressed companies from the non-distressed ones. For example, an enterprise, which has a high systematic risk corresponding to macroeconomic conditions, e.g., growth of industrial production, inflation, changes in interest rates, and changes in money supply (M2), may be exposed to the possibility of financial distress. In this article, the multi-factor model represents the relationship between the firm’s stock return and the firm’s sensitivity to macroeconomic factors. In other words, changes in macroeconomic factors and the firm’s sensitivity to those changes will affect the firm’s stock return. Then, the relationship between the firm’s probability of financial distress and the firm’s stock return is such that a firm, which encounters a high return, will have a low probability of financial distress. This process needs a two-step analysis whereby changes in macroeconomic factors and the firm’s sensitivity to those factors affect the firm’s stock return, with a firm’s stock return in turn having an effect on the firm’s probabilityof financial distress. Therefore, the two- step logit can be applied as used in Maddala (1986). This line of methodology has also been used in Thomson (1992) and Theodossiou et al. (1996). For example, the study by Thomson (1992) uses two-step
  • 8. MultinationalFinanceJournal110 TABLE 1. Sample Characteristics Description Full Sample Distressed Non-distressed Total assets 19,772,150.10 21,690,440.67 19,462,748.39 Earnings before interest and taxes 512,615.58 –84,358.06 608,901.65 Total equity 3,371,051.44 2,683,971.35 3,481,870.81 Total debt 16,401,098.66 19,006,469.32 15,980,877.58 Number of observations 396 55 341 Note: The figures are expressed in thousand baht. As of December 31, 1996, the exchange rate between the US$ and the Thai baht is US$ 1 = 25.7 baht. However, after the Ministry of Finance of Thailand changed the exchange rate regime from the currency basket to the managed float system on July 2, 1997, the currency exchange of US$ 1 as of December 31, 1997, was equal to 48.15 baht.
  • 9. Investigation of Financial Distress for Thai Firms 111 logit regression approach as a link between official bank failure and bank insolvency. Theodossiou et al (1996) also reexamine the economic factors determining the decision to acquire a financially distressed firm using a sequential response logit (SRL) model. In their model,acquisition,nonacquisition,andwealthyoutcomesareexpressed as a two-stage sequence of binary outcomes, healthy/distress in stage one, and acquisition/non-acquisition in stage two. Therefore, logistic discriminant analysis used in this article will apply this two-step idea in order to estimate return on a firm’s stock, and then the estimated stock return will be used as a representation of a macroeconomic sensitivity indicating a firm’s financial distress. The estimated stock return calculated from the multi-factor model is designed as a link to the micro-crisis model. In other words, the firm’s estimated return calculated from the multi-factor (or macroeconomic factors) model represents the macro effects incorporated into the micro- crisis prediction model. Several researchers, e.g., Beaver (1966) and Scott (1981), confirm the relationship between a firm’s probability of bankruptcy and the firm’s stock return. But because the firm’s actual stock return reflects both the systematic risk and firm-specific risk, the motivation of the two-step is to capture only the systematic risk in which the economic crisis has affected the firms in the economy as a whole. In general, the micro-crisis model and the related macro factors imply the following specifications (the indirect test): ,prob Y Z i i = = + − 1 1 1 0 5 0 5exp where , (1), ˆ i i j i j Z a bR c= + + ∑ ,$ $ , ,R Fi k i k k = + ∑β β0 Yi is assigned the value of 1 if firm i is financially distressed and the value of zero otherwise, Prob(Yi=1) is the probability with which firm i is classified as a financially distressed firm, Zi is a linear function in which a, b, and c are coefficients of variables, Xj,i is the financial
  • 10. Multinational Finance Journal112 characteristic j of firm i, is the estimated macroeconomic factor k,$Fk and is the estimated monthly stock return of firm i, specified as a$Ri linear function of the macroeconomic factor .$Fk The probability, Prob (Yi=1), depends on the logistic function of a linear relationship of a set of variables including and Xi,j. The$Ri estimated monthly stock return of the firm i, , can be calculated from$Ri amulti-factormodelinwhichtheestimatedmacroeconomicfactors, ,$Fk are used to develop the model. The relationship between a firm’s probability of financial distress and its stock return has been pointed out by Beaver (1968) and Scott (1981). While the first study showed the decline of a firm’s stock return as the firm approached failure, the second confirmed that the probability of bankruptcy depended on the firm’s stock return. Therefore, a firm’s stock return is a good reflection of the market expectation of the firm’s probability of financial distress. The actual stock return, Ri , however, represents both systematic risk andfirm-specificrisk;e.g., ,inwhichFrepresentsR F ei i i i= + +β β0 1, , the systematic factor(s) and ei contains the firm-specific risk. Since the firm’s financial characteristics, discussed later, are related to the firm- specific risk, the use of the actual stock return, together with the firm’s financial characteristics, as explanatory variables to the model will be a double consideration of the firm-specific risk. In addition, the 1997 economic crisis has systematically hit the overall listed firms, so the motivation of this study is an attempt to find a proxy for systematic risk corresponding to the firm’s bankruptcy risk. The estimation specification of will carve out the firm-specific risk and capture only$Ri the systematic risk of the firm. Then the use and financial$Ri characteristics combines the systematic and firm-specific risks as the representationofexplanatoryvariablestothemacro-relatedmicro-crisis model without a redundancy of the firm-specific risk consideration. C. Financial variables The set of variables, Xj,i, represents financial characteristics of the firm i. Shivaswamy,Hoban,andMatsumoto(1993)studiedthirteenresearch papers and summarized the frequency of financial ratios used in those papers. The most frequently used ratios were: current ratio, leverage ratio, and profitability ratio. Altman and Altman et al. (1968, 1983, 1984, 1994) developed numerous models predicting bankruptcy and financial distress. In particular, Altman (1984) surveyed studies on
  • 11. Investigation of Financial Distress for Thai Firms 113 businessfailuremodelsinvarious countries and reviewedtheutilization of variable profiles employed in those studies. The most frequently used financial ratios were: working capital to total assets, retained earnings to total assets, earnings before interest and taxes to total assets, market value of equity to total liabilities, and sales to total assets. However, the literature in this area of finance and accounting shows that researchers usually base their models on the capital, assets, management, earnings, and liquidity of a firm, which are the so-called CAMEL categories. FollowingSalchenberger, Cinar, andLash (1992), the financial ratios are categorized only from each element of the CAMEL framework. The set of variables, Xj, following the CAMEL categories is selected as follows: Capital: X1 = book value of stockholders’ equity / total assets (SETA) Assets: X2 = retained earnings / total assets (RETA) Management and Earnings: X3 = operating income / net sales (OINS) Liquidity: X4 = net working capital / total assets (WCTA) These four variables will be used as the financial characteristics of the firms in the sample and as explanatory factors to the model. The dependence on a set of financial ratios as a sole explanatory set of variables, however, has raised a question of whether they are sufficient to explain a firm’s financial situation when the crisis strikes the economy as a whole. The reason is that a firm’s accounting system is prone to accounting standard, e.g., Generally Accepted Accounting Principles (GAAP), and it may not represent a good rationale for a firm’s financial distress due to the economic slump. So a firm’s sensitivitytomacroeconomicfactorsshouldbeincluded,as explanatory variables to the firm’s financial distress and the estimated stock return whichincorporatesthe effect of the firm’s sensitivitytomacroeconomic factors should be a good representation of the macroeconomic factor for explaining the firm’s probability of financial distress. D. Multi-factor model Demirguc-Kunt and Detragiache (1997) use ten variables as macroeconomic factors in their logit model. However, only four variables (real quarterly GDP growth, real short-term interest rate, inflation,andM2moneysupplytointernationalreserves)aresignificant variables, which can explain the probability of banking crises. These four macroeconomic factors have also been widely used
  • 12. Multinational Finance Journal114 econometrically. For the sample of Thai economic conditions, monthly growth of the production-manufacturing index will provide more observations than will the use of quarterly real GDP growth. Therefore, the macro-factors, Fk, are selected as follows: F1 = monthly growth of the production manufacturing index (PMI) F2 = monthly inflation or changes in the consumer price index (CPI) F3 = monthly changes in interest rates (INT) F4 = monthly changes in M2 money supply (MS2) The estimated changes in the above economic conditions will be used to calculate the estimated stock returns of firm i according to the following equation: ,$ $ $ $ $ , , , , ,R F F F Fi i i i i i= + + + +β β β β β0 1 1 2 2 3 3 4 4 where 0,i is the intercept and 1,i, 2,i, 3,i, and 4,i are coefficients of variables representing the systematic risks or the sensitivity of firm i to the changes in industrial production, consumer price index, interest rates, and M2 money supply, respectively. It should be noted that for the specification in the indirect test (equation 1), the estimated stock return is a complete reflection of the sensitivity of the firm to macroeconomic conditions as well as the macroeconomic conditions. Alternatively, the effect of a firm’s sensitivity of macroeconomic factors on the probability of the firm’s financial distress can be specified as follows (the direct test): prob Y Z i i = = + − 1 1 1 0 5 0 5exp where (2) .Z a b c X ei k k i j j i i jk = + + +∑∑ β , , The above model tests the effect of the sensitivity of firms in terms of financial distress to four macroeconomic conditions: growth of industrial production, inflation, changes in interest rates, and changes in M2 money supply. In this direct test, the explanatory factors of the firm’s probability of financial distress include the firm’s sensitivity to macroeconomic factors and the firm’s financial characteristics.
  • 13. Investigation of Financial Distress for Thai Firms 115 11. The monthly data of the stock return and the macroeconomic factors were collected startingfromJanuary1987 toDecember1996. Therefore,thereare120monthlyobservations that are employed in regressing the stock return over the macroeconomic factors. The systematic risks, k,i, of firm i are obtained from the regression of the monthly stock returns of the firm i at time ti, Ri,t on the monthly macroeconomic factors at time ti, Fj,t.11 The equation is such that: ,R F F F F ei i i t i t i t i t i t= + + + + +β β β β β0 1 1 2 2 3 3 4 4, , , , , , , , , , where ei,t, the error term, represents the firm-specific risk of the firm i at time t. IV. Empirical Results The descriptive statistics of variables are reported in table 2, while the resultsofthemacro-relatedmicro-crisispredictionmodels,bothindirect and direct tests, are presented in table 3. The results indicate that the estimated stock return variable corresponding to the estimated changes in macroeconomic factors is significantly different from zero. The negative sign suggests that a firm with a higher estimated rate of return has a lower probability of financial distress. The rationale has been pointed out by Beaver (1966) as the declining of a firm’s stock return signaling the firm’s vulnerability to the possibility of bankruptcy. The significance of this variable (at the 5% significance level) also indicates the importance of macroeconomic conditions and those conditions can strongly influence the probability of the firm’s financial distress. The results of the alternative direct specification (equation 2) are also reported in table 3. The results indicate that the only significant macro factor is the sensitivity of the firms to inflation or changes in the consumer price index. In other words, only the systematic risk of a firm exposed to inflation affects the probability of the firm’s financial distress. The sign suggests that the more a firm faces inflation risk, the higher the firm’s vulnerability to the risk of financial distress. For both alternative models (the indirect test in equation 1 and the direct test in equation 2), the financial characteristics such as stockholders’ equity to total assets (SETA) and working capital to total assets (WCTA) are significant at the 1% significance level. The relationship is such that the greater the SETA and WCTA ratios, the
  • 14. MultinationalFinanceJournal116 TABLE 2. Descriptive Statistics of Variables for the Full, the Non-distressed, and the Distressed Samples Full Sample (396 Firms) Non-distressed Sample (341 Firms) Distressed Sample (55 Firms) Model Mean Variable Max Min Mean Max Min Mean Max Min Mean Difference A. Macroeconomic Factor Ri 3.5138 –11.5892 –.2962 3.5138 –2.5304 –.2214 .9808 –11.5892 –.7602 .5389 [.76] (2.3)** (4.16)*** {6.43}*** PMI 5.6707 –1.441 .0374 1.6086 –1.441 .0243 5.6707 –.8706 .1185 –.0942 [.48] (–.82) (–1.34) {4.45}*** CPI 26.303 –17.5141 1.1273 17.8946 –17.5141 .6371 26.303 –6.9741 4.166 –3.5288 [5.1] (–3.8)*** (–4.89)*** {2.03}*** int .4366 –1.6489 –.0664 .4366 –.5001 –.0586 .1775 –1.6489 –.1143 .0557 [.14] (1.65) (2.67)*** {4.19}*** MS2 26.6999 –18.5894 1.5073 11.7656 –18.5894 1.314 26.6999 –5.3813 2.706 –1.3921 [3.42] (–2.05)** (–2.82)*** {2.49}***
  • 15. InvestigationofFinancialDistressforThaiFirms117 TABLE 2. (Continued) B. Financial Characteristic SETA 1.003 –2.1488 .4201 1.0030 –2.1488 .4601 .7051 –1.1681 .1718 .2883 [.24] (7.26)*** (7.41)*** {1.06} RETA 1.3293 –3.0053 .1007 1.3293 –3.0053 .1331 .4259 –2.9265 –.098 .2308 [.30] (3.13)*** (4.6)*** {3.06}*** OINS 7.195 –17.1278 .0893 7.195 –17.1278 .1193 .42 –3.4021 –.0965 .2158 [1.1] (1.83)* (1.29) {2.76}*** WCTA 5.3238 –44.2817 –.405 5.3238 –44.2817 –.2145 .3336 –7.5143 –1.5864 1.3719 [2.43] (4.64)*** (3.86)*** {1.67}** Note: Ri= 0,i+ 1,iF1,t+ 2,iF2,t+ 3,iF3,t+ 4,iF4,t +ei,t The macroeconomic variables are from monthly data during 1987-1996. á i is the estimated monthly stock return of the firm i. PMI represents the systematic risks or the sensitivity of a firm to the changes in industrial production. CPI represents the systematic risks or the sensitivity of a firm to the changes in consumer price index. INT represents the systematic risks or the sensitivity of a firm to the changes in interest rates. MS2 represents the systematic risks or the sensitivity of a firm to the changes in M2 money supply. SETA represents the ratio of stockholders’equity to total assets. RETA represents the ratio of retained earnings to total assets. OINS represents the ratio of operating income to net sales. WCTA represents the ratio of working capital to total assets. Parentheses include the standard errors in [ ], t-values, for H0: variances are equal, in ( ), and F-values, for H0: means are not significantly different from zero, in { }. ***, **, and * denote statistical significance at the .01, .05, and .1 levels, respectively.
  • 16. Multinational Finance Journal118 TABLE 3. The Macro-related Micro-crisis Prediction Models: Estimated Parameters and Wald Chi-square Statistics in Parentheses Estimates of Estimates of Coefficients Elasticitya Coefficients Elasticityb Constant –.9036 n/a –1.1033 n/a (8.51)*** (11.32)*** Macroeconomic Factor –.503 .3014 (5.69)** PMI –.1288 .0795 (.12) CPI .0986 .4435 (8.7)*** INT –1.5497 .1695 (1.25) MS2 –.0211 .0708 (.15) Financial Characteristic SETA –3.5065 .7899 –3.1373 .6993 (18.64)*** (14.66)*** RETA –.3395 .0675 –.5313 .0038 (.28) (.65) OINS .163 .1874 .1077 .0129 (.52) (.39) WCTA –.1228 .2046 –.1171 .1985 (7.12)*** (6.8)*** –2 Log L 259.070 253.330 Model Chi-Square 60.06*** 65.80*** Note: For model 1 , where . Forprob Y Z i i = = + − 1 1 1 0 5 0 5exp Z a bR c Xi i j j i j = + +∑$ , model 2 , where .prob Y Z i i = = + − 1 1 1 0 5 0 5exp Z a b c X ei k k i j j i i jk = + + +∑∑ β , , is the estimated monthly stock return of the firm i. PMI represents the systematic risks or $ Ri the sensitivity of a firm to the changes in industrial production . CPI represents the systematic risks or the sensitivity of a firm to the changes in consumer price index. INT represents the systematic risks or the sensitivity of a firm to the changes in interest rates. MS2 represents the systematic risks or the sensitivity of a firm to the changes in M2 money supply. SETA represents the ratio of stockholders’ equity to total assets. RETA represents the ratio of retained earnings to total assets. OINS represents the ratio of operating income to net sales. WCTA represents the ratio of working capital to total assets. Parentheses include the standard errors in [ ], t-values, for H0: variances are equal, in ( ), and F-values, for H0: means are not significantly different from zero, in { }. ***, **, and * denote statistical significance at the
  • 17. Investigation of Financial Distress for Thai Firms 119 TABLE 4. The Full-sample Test for Model 1 Distress Prediction Non-distressed Prediction Total Prediction Accuracy Accuracy Accuracy Cutoff Point Amount % Amount % Amount % 0 55 100.00% 0 0.00% 55 13.89% .1 48 87.27% 205 60.12% 253 63.89% .2 33 60.00% 294 86.22% 327 82.58% .3 21 38.18% 323 97.95% 344 86.87% .4 13 23.64% 334 97.95% 347 87.63% .5 8 14.55% 336 98.53% 344 86.87% .6 6 10.91% 338 99.12% 344 86.87% .7 4 7.27% 339 99.41% 343 86.62% .8 3 5.45% 339 99.41% 342 86.36% .9 2 3.64% 339 99.41% 341 86.11% 1 0 0.00% 341 100.00% 341 86.11% Note: For model 1 , where ,prob Y Z i i = = + − 1 1 1 0 5 0 5exp , ˆ i i j j iZ a bR c X= + ∑ . The sample consists of 396 firms in which 341 firms are non-0, , ˆ ˆ i i k iR Fβ β= + ∑ distressed and 55 firms are financially distressed in 1997. TABLE 5. The In-sample Test for Model 1 Distress Prediction Non-distressed Prediction Total Prediction Accuracy Accuracy Accuracy Cutoff Point Amount % Amount % Amount % 0 30 100.00% 0 0.00% 30 12.99% .1 25 83.33% 154 76.62% 179 77.49% .2 15 50.00% 185 92.04% 200 86.58% .3 15 50.00% 195 97.02% 210 90.91% .4 12 40.00% 199 99.01% 211 91.34% .5 10 33.33% 199 99.01% 209 90.48% .6 10 33.33% 199 99.01% 209 90.48% .7 6 20.00% 199 99.01% 205 88.75% .8 5 16.67% 199 99.01% 204 88.31% .9 4 13.33% 199 99.01% 203 87.88% 1 0 0.00% 201 100.00% 201 87.01% Note: For model 1 , where ,prob Y Z i i = = + − 1 1 1 0 5 0 5exp , ˆ i i j j iZ a bR c X= + ∑ . The sample is divided into two subsamples: 60% of the full sample as0, , ˆ ˆ i i k iR Fβ β= + ∑ the in-sample (learning sample) and 40% of the full sample as the out-sample (holdout sample).
  • 18. Multinational Finance Journal120 TABLE 6. The Out-sample Test for Model 1 Distress Prediction Non-distressed Prediction Total Prediction Accuracy Accuracy Accuracy Cutoff Point Amount % Amount % Amount % 0 25 100% 0 0.0% 25 15.15% .1 22 88% 103 73.57% 125 75.76% .2 18 72% 120 85.71% 138 83.64% .3 13 52% 126 90.00% 139 84.24% .4 11 44% 133 95.00% 144 87.27% .5 11 44% 135 96.43% 146 88.49% .6 5 20% 136 97.14% 141 85.46% .7 3 12% 137 97.86% 140 84.85% .8 2 8% 137 97.86% 139 84.24% .9 2 8% 137 97.86% 139 84.24% 1 0 0% 140 100.00% 140 84.85% Note: For model 1 , where ,prob Y Z i i = = + − 1 1 1 0 5 0 5exp , ˆ i i j j i Z a bR c X= + ∑ . The sample is divided into two subsamples: 60% of the full sample as0 , , ˆ ˆ i i k i R Fβ β= + ∑ the in-sample (learning sample) and 40% of the full sample as the out-sample (holdout sample). TABLE 7. The Full-sample Test for Model 2 Distress Prediction Non-distressed Prediction Total Prediction Accuracy Accuracy Accuracy Cutoff Point Amount % Amount % Amount % 0 25 100% 0 0.00% 25 15.15% .1 22 88% 103 73.57% 125 75.76% .2 18 72% 120 85.71% 138 83.64% .3 13 52% 126 90.00% 139 84.24% .4 11 44% 133 95.00% 144 87.27% .5 11 44% 135 96.43% 146 88.49% .6 5 20% 136 97.14% 141 85.46% .7 3 12% 137 97.86% 140 84.85% .8 2 8% 137 97.86% 139 84.24% .9 2 8% 137 97.86% 139 84.24% 1 0 0% 140 100.00% 140 84.85% Note: For model 2 , where,prob Y Z i i = = + − 1 1 1 0 5 0 5exp , , .i k k i j j i i k j Z a b c X eβ= + + +∑ ∑ The sample consists of 396 firms in which 341 firms are non-distressed and 55 firms are financially distressed in 1997.
  • 19. Investigation of Financial Distress for Thai Firms 121 TABLE 9. The Out-sample Test for Model 2 Distress Prediction Non-distressed Prediction Total Prediction Accuracy Accuracy Accuracy Cutoff Point Amount % Amount % Amount % 0 30 100.00% 0 0.00% 30 12.99% .1 26 86.67% 153 76.12% 179 77.49% .2 17 56.67% 183 91.05% 200 86.58% .3 15 50.00% 195 97.02% 210 90.91% .4 13 43.33% 199 99.01% 212 91.78% .5 11 36.67% 199 99.01% 210 90.91% .6 9 30.00% 199 99.01% 208 90.04% .7 8 26.67% 199 99.01% 207 89.61% .8 5 16.67% 199 99.01% 204 88.31% .9 4 13.33% 200 99.50% 204 88.31% 1 0 0.00% 201 100.00% 201 87.01% Note: For model 2 , whereprob Y Z i i = = + − 1 1 1 0 5 0 5exp , , .i k k i j j i i k j Z a b c X eβ= + + +∑ ∑ The sample is divided into two subsamples: 60% of the full sample as the in-sample (learning sample) and 40% of the full sample as the out-sample (holdout sample). TABLE 8. The In-sample Test for Model 2 Distress Prediction Non-distressed Prediction Total Prediction Accuracy Accuracy Accuracy Cutoff Point Amount % Amount % Amount % 0 30 100.00% 0 0.00% 30 12.99% .1 26 86.67% 153 76.12% 179 77.49% .2 17 56.67% 183 91.05% 200 86.58% .3 15 50.00% 195 97.02% 210 90.91% .4 13 43.33% 199 99.01% 212 91.78% .5 11 36.67% 199 99.01% 210 90.91% .6 9 30.00% 199 99.01% 208 90.04% .7 8 26.67% 199 99.01% 207 89.61% .8 5 16.67% 199 99.01% 204 88.31% .9 4 13.33% 200 99.50% 204 88.31% 1 0 0.00% 201 100.00% 201 87.01% Note: For model 2 , whereprob Y Z i i = = + − 1 1 1 0 5 0 5exp , , .i k k i j j i i k j Z a b c X eβ= + + +∑ ∑ The sample is divided into two subsamples: 60% of the full sample as the in-sample (learning sample) and 40% of the full sample as the out-sample (holdout sample).
  • 20. Multinational Finance Journal122 12. Unlike the logit coefficients, the elasticity is independent of measurement units for the variables. 13. Type I errors is the probability of a failed firm misclassified as an unfailed firm, and type II error is the probability of an unfailed firm misclassified as a failed firm. can be defined as the misclassification of the failed firm into the unfailed firm and type II errors can be defined as the misclassification of the unfailed firm into the failed firm. lower the probability of financial distress. Also, table 3 reports the elasticities, as defined in Theodossiou, et al. (1996), which measure the importance of the variables in the model.12 It can be seen that SETA and , for model 1, and SETA and bCPI , for model 2, are the most$Ri important variables, consistent with the Wald chi-square statistics. Tables 4-9 show the predictive ability of the macro-related micro- crisis prediction models at various cut-off points. Since type I errors are more costly than type II errors13 , the study attempts to determine a cut- off point that correctly discriminates the distressed firms at a greater number than does the non-distressed firms or, in other words, in which type I errors are approximately smaller than type II errors. On average, the cut-off point at .1 seems to be optimal, indicating that a firm that has a probability of financial distress of 10% or over will be classified as a distressed firm and a firm that has a probability under 10% will be classified as a non-distressed firm. The prediction results for the indirect model (equation 1) are shown in table 4-6. At the cut-off point of .1, the total prediction accuracy equals to 63.89% for the full-sample test, 77.49% for the in-sample test, and 75.76% for the out-sample test as shown in table 4-6. The prediction results for the direct model (equation 2) are shown in table 7-9. At the cut-off point of .1, the total prediction accuracy equals to 68.43% for the full-sample test, 77.49% for the in-sample test, and 70.91% for the out-sample test as shown in table 7-9. The out-sample predictive ability of over 70% for both models signifies promising outcomes of this study. V. Conclusions and Discussion This article employed the 1997 economic crisis in Thailand as a case study. Using logit analysis, the model attempts to portray corporate bankruptcy caused by economic collapse as systematic risk. The macro- related micro-crisis exploration model developed in this study has two specifications. The first specification (the indirect test) uses firms’
  • 21. Investigation of Financial Distress for Thai Firms 123 estimated stock returns calculated from the estimated changes in macroeconomic variables and firms’ sensitivities to those macro variables as proxies of macro-related factors, while the second specification (the direct test) uses firms’ sensitivities to macro variables as proxies for macroeconomic factors. The contribution of the model is to bridge a firm’s sensitivity to macroeconomic conditions and its financial characteristics in order to investigate the firm’s financial distress. The results are promising and confirm that macroeconomic conditions are important factors in determining a firm’s possibility of financial crises. The estimated stock return variable shows significance in the indirect-test specification (equation 1) and the sensitivity-to- inflation variable is also significant at the .01 level in the direct-test specification (equation 2). Since 1996-1997, the Thai economy showed signs of a steep decline causing negative returns for equity investors. The empirical findings demonstrate that, in an economic crisis, the higher or less negative a firm’s return, the lower the probability of the firm’s financial distress. Inflation is the most crucial variable because the higher a firm’s sensitivity to inflation, the more likely the firm’s exposure to financial distress. For testing the predictive ability of the model, the study uses the cut- off point at .1 to separate financially distressed firms from non- distressed firms because type I errors are quite costly, relative to type II errors. For both specifications, the predictive ability or accuracy ranges between 60-80% of the sample. An out-sample test of predictive ability confirms that the model has an accuracy of over 70% of the holdout sample. In conclusion, the article develops the investigation model using the 1997 Thailand financial crisis. The model incoporates the firms’ sensitivitytomacroeconomicvariablesandtheirfinancialcharacteristic. It will be useful in providing warning signals of upcoming crises so that policy makers can seek protective measure in order to immunize the economy and protect against contagious, potentially lethal financial diseases. References Alba, P.; Claessens, S.; and Djankov, S. 1998. Thailand’s Corporate financing and governance structures: Impact on firms’ competitiveness. Conference
  • 22. Multinational Finance Journal124 onthailand’sdynamiceconomicrecoveryandcompetitiveness.May20-21, 1998, UNCC, Bangkok, Thailand. Altman, E. I. 1968. Financial ratios, discriminant analysis and the prediction of corporatebankruptcy.TheJournalofFinance23(September):589-609. Altman, E. I. 1983. Multidimensional graphics and bankruptcy prediction: a comment. Journal of Accounting Research 21(Spring): 297-299. Altman, E. I. 1984. The Success of business failure prediction models: an international survey. Journal of Banking and Finance 8: 171-198. Altman, E. I. 1994. Corporate distress diagnosis: Comparisons using linear discriminant analysis and neural network (the Italian experience). Journal of Banking and Finance 18: 505-529. Altman, E. I. and Saunders, A. 1998. Credit risk measurement: Developments over the last 20 years. Journal of Banking and Finance 21: 1721-1742. Barnes, P. 1987. The analysis and use of financial ratios: A review article. Journal of Business Finance and Accounting 14: 449-461. Beaver, W. H. 1966. Financial ratios as predictors of failure. Empirical research in accounting: selected studies. Journal of Accounting Research 4 (Supplement). Caprio, G. and Klingebiel, D. 1996. Dealing with bank insolvencies: Cross countryexperience, The World Bank, Washington, D.C. Corsetti, G.; Pesenti, P.; and Roubini N. 1998. What Caused the Asian currency and financial crisis? Unpublished Manuscript. Demirguc-Kunt, A. and Detragiache, E. 1997. The determinants of banking crises: Evidence from developing and developed countries. International Monetary Fund Working Paper. Eichengreen, B.; Rose, A. K.; and Wyplosz, C. 1996. Contagious currency crises. NBER Working Paper No. 5681. Eichengreen, B. and Rose, A. K. 1998. Staying Afloat when the wind shifts: externalfactorsandemerging-marketbankingcrises.NBERWorkingPaper No.6370. Flagg, J. C.; Giroux, G. A.; and Wiggins, C. E. Jr. 1991. Predicting corporate bankruptcy using failing firms. Review of Financial Economics 1: 67-78. Flood, R. and Marion, N. 1998. Perspectives on the Recent Currency Crisis Literature. NBER Working Paper No. 6380. Gilbert, L. R.; Krishnagopal, M.; and Schwartz, K. B. 1990. Predicting bankruptcy for firms in financial distress. Journal of Business Finance and Accounting 17 (Spring):161-171. Glick, R. and Rose, A. K. 1998. Contagion and trade: Why are currency crises regional? NBER Working Paper No. 6806. Hopwood, W.; McKeown , J. C. ; and Mutchler, J. F. 1989. A Test of the incremental explanatory power of opinions qualified for consistency and uncertainty. The Accounting Review 66:28-48. Johnsen, T. and Melicher, R. W. 1994. Predicting Corporate bankruptcy and financial distress: information value added by multinomial ligit models. Journal of Economics & Business 46:269-286.
  • 23. Investigation of Financial Distress for Thai Firms 125 Kaminsky, G. and Reinhart, C. M. 1996. The Twin crises: The causes of banking and balance-of-payment problems. Federal reserve board and international monetary fund. Unpublished Manuscript. Kaminsky, G.; Lizondo, S.; and Reinhart, C. M. 1997. Leading indicators of currency crises. International monetary fund. Working Paper. La Porta, R.; Lopez-de-Silanes, F. ; Shleifer, A.; and Vishny, R. W. 1998. Law and finance. Journal of Political Economy, forthcoming. Maddala, G. S. 1986. Econometric issues in the empirical analysis of thrift institutions’ insolvency and failure. Invited Research Working Paper 56, Federal Home Loan Bank Board, October. Mensah, Y. M. 1983. The differential bankruptcypredictive abilityof specific price level adjustments: Some empirical evidence. The Accounting Review LVIII(April):228-246. Ohlson, J. A. 1980. Financial ratios and the probabilistic prediction of bankruptcy. Journal of Accounting Research 18(Spring):109-130. Platt, H. E. and Platt, M. B. 1990. Improving ex ante bankruptcy classification using stable predictive variables. Journal of Business Finance and Accounting 17:31-51. Platt, H. E. and Platt, M. B. 1991. A note on the use of industry-relative ratios in bankruptcy prediction. Journal of Banking and Finance 15:1183-1194. Poston, K. M.; Harmon, W. K.; and Gramlich, J. D. 1994. A test of financial ratios as predictors of turnaround versus failure among finanially distressed firms. Journal of Applied Business Research 10: 41-56. Rachel, S. and Sachs, J. 1998. The onset of the east Asian financial crisis. NBER Working Paper No. 6680. Rose, R. S.; Andrews, W. T.; and Giroux, G. A. 1982. Predicting business failure: A macroeconomic perspective. Journal of Accounting, Auditing, and Finance (Fall): 20-31. Salchenberger, L. M.; Cinar, E. M.; and Lash, N. A. 1992. Neural networks: A new tool for predicting thrift failures. Decision Sciences 23: 899-916. Scott, J. 1981. The probability of bankruptcy: A comparison of empirical prediction and theoretical model. Journal of Banking and Finance 5: 317- 344. Shivaswamy, M.; Hoban, J. P.; and Matsumoto, K. 1993. A behavioral analysis of financial ratios. The Mid-Atlantic Journal of Business 29 (March) :7-24. Theodossiou, P.; Kahya, E. ; Saidi, R.; and Philippatos, G. 1996. Financial distress and corporate acquisitions: Further empirical evidence. Journal of Business Finance and Accounting 23 (July): 699-719. Thomson, J. B. 1992. Modelingthe bank regulator’s closure option: A two-step logit regression approach. Journal of Financial Services Research. Zavgren, C. 1983. The prediction of corporate failure: The state of the art. Journal of Accounting Literature 2: 1-35. Zmijewski, M. E. 1984. Methodological issues related to the estimation of financial distress prediction models. Journal of Accounting Research 22 (Supplement): 59-85.