The Use Of Ohlson's O-Score For Bankruptcy Prediction In Thailand

The Journal of Applied Business Research – November/December 2015 Volume 31, Number 6
Copyright by author(s); CC-BY 2069 The Clute Institute
The Use Of Ohlson's O-Score
For Bankruptcy Prediction In Thailand
Judy Ramage Lawrence, Christian Brothers University, USA
Surapol Pongsatat, Ramkhamhaeng University, Thailand
Howard Lawrence, University of Mississippi, USA
ABSTRACT
Business failure is a major concern to all parties involved and can create high costs and heavy
losses. If bankruptcy can be predicted with reasonable accuracy ahead of time, firms can better
protect their businesses and can take action to minimize risk and loss of business, and perhaps
even prevent the bankruptcy itself. Bankruptcy prediction in thailand is important because
business in thailand has historically operated on a system of trust where one person doing
business trusts the other to perform as agreed upon in written and oral contracts. The threat of
bankruptcy tends to diminish that trust and weakens the country's ability to prosper. While
research in bankruptcy has been extensive, there has been only limited research on bankruptcy
prediction in thailand. This study expands on an earlier study by pongsatat, et al (1994) using
ohlson's o-score to determine if there a significant difference in ohlson’s o-score as measured by
ohlson’s logit analysis model between bankrupt and non-bankrupt firms in thailand. The results
of the independent samples t-test demonstrates that there are significant differences in the
population means for one year, two years and three years prior to bankruptcy at the 0.05 level.
Therefore the null hypothesis that there is no difference in the mean of ohlson’s o-score as
measured by logit analysis between bankrupt and non-bankrupt firms in thailand is rejected.
Keywords: Bankruptcy; Prediction; Ohlson O-Score; Thailand
INTRODUCTION
he failure of a business organization has significant economic effects for its owners, creditors, and to
society overall. Almost every firm has debt and in the majority of cases that debt is the result of good
financial planning, because debt financing is often a lower cost alternative to equity financing.
Unfortunately, with debt there comes risk, and the ultimate risk is that of bankruptcy. To avoid this problem firms
need reliable ways to predict bankruptcy. This is so because with proper warning, businesses can often take the
necessary measures to minimize this risk and possibly prevent the bankruptcy itself.
Thailand is a country that has historically operated on a system of personal trust between parties. Oral
contracts are more common in Thailand than in countries such as the United States. Bankruptcy causes that trust to
be diminished and ultimately hurts the country’s ability to prosper. There have been many studies of bankruptcy in
Asia with the majority finding that the causes of bankruptcy in Asian countries can be quite different from studies in
western countries (Pongsatat, et al, 2004: Evans, et al, 2013; Sirirattanaphonkun & Pattarathammas, 2012;
Treewichayapong, et al 2011; Reynolds, et al, 2002; Urapeepatanapong, et al, 1998). The analysis of financial ratios
has been the primary method used for analysis. (See: Pongsatat, et al, 2004; Clark & Jung, 2002; Brietzke, 2001;
Pugh & Dehesh, 2001; Haley, 2000; & Tirapat & Nittayagasetwat, 1999). Four of these studies addressed
bankruptcy in Thailand (Pongsatat, et al, 2004; Evans, et al, 2013; Reynolds, et al, 2002 & Tirapat &
Nittayagasetwat, 1999). The remaining studies addressed bankruptcy in South Korea (Clark & Jung, 2002; Haley,
2000) Indonesia (Brietzke, 2001), and Taiwan (Clark & Jung, 2002). As in the west, financial strength has been the
primary indicator of the ability to avoid bankruptcy. Other factors have also come into play, with the inability to
adjust to changing environmental factors being the major cause. Such factors as customer preferences, social norms,
new legal requirements and rapidly changing global competitive forces have meant that companies must adjust or
T

The Journal of Applied Business Research – November/December 2015 Volume 31, Number 6
Copyright by author(s); CC-BY 2070 The Clute Institute
die. As these forces continue, companies must find better ways to predict bankruptcy. This is all the more
important in Thailand today given that much of the investment that previously came to Thailand now goes to
countries such as China and Vietnam.
Two of the primary methods used in the west to predict bankruptcy are Altman's Z-score and Ohlson's O-
score. Pongsatat, et al, (2004) examined the comparative ability of Ohlson’s Logit model and Altman’s four-
variance model for predicting bankruptcy in Thailand to determine if there was a difference in their ability to predict
bankruptcy in Thailand. This study expands on Pongsatat’s 2004 study to specifically examine Ohlson’s logit
model’s ability to predict bankruptcy for both financial and non-financial firms.
Pongsatat’s 2004 study found that this analysis is important because evidence suggests that studies
conducted in countries such as the united states and the european union may not have relevance to a country such as
thailand. Pongsatat cited tirapat and nittayagasetwat (1999) who found that the institutional structure of the financial
sector in thailand is characterized as a bank centered financial system, while that of the u.s. is a market based
system. For example, tirapat and nittayagasetwat (1999) point out that in thailand, the credits from financial
institutions such as commercial banks and finance companies can be five or more times larger than the funds raised
from the stock and bond markets. In 1997, at the height of the asian crisis, the amount was 16.7 times higher. These
are significantly larger numbers than what could be found in most western economies. Other differences in thailand
cited by tirapat and nittayagasetwat (1999) included a higher concentration of shares being held by a few individuals
and more shares with large family ownership.
DISCUSSION
The cause of business failures is a matter that has been discussed and researched at length and it has been a
topic of interest for hundreds of years. Throughout these years, ratio analysis has been at the heart of the matter.
According to Horrigan (1968), the evolution of ratio analysis came about in approximately 300 BC, primarily
because of Euclid’s analysis of the properties of ratios in Book V of his Elements. Wall (1919) may have been the
first to recognize that a relative ratio criteria could be used in place of the more popular absolute ratio criteria.
Horrigan (1968) described Wall’s Study of Credit Barometrics, as historic “because it was a widely-read overt
departure from the customary usage of a single ratio with an absolute criterion” (p. 285–286). According to
Horrigan, Wall studied 981 firms and used seven ratios in his analysis. The firms were stratified by industry and
geographical areas with nine different subdivisions in each stratum. In doing this, “Wall...had, in effect,
popularized the ideas of using many ratios and using empirically determined relative ratio criteria” (Horrigan, 1968,
p. 286).
Fitzpatrick (1931) also conducted early research in this area. Fitzpatrick examined 13 ratios to see if they
could indicate failure. He examined each ratio individually using a univariate analysis. Unfortunately, Fitzpatrick
was not able to show a significant relationship with bankruptcy. Thirty-five years later Beaver (1966a and 1966b)
conducted a univariate analysis and was able to show a significant relationship. Beaver’s study is considered one of
the classics in the field.
Altman (1968) moved from using a univariate approach to using multiple discriminant analysis to predict
bankruptcy. Even today, researchers still regard Altman’s Z-Score as a good indicator of a company’s ability to
avoid bankruptcy. Altman later revised his model to incorporate a “four variable Z-Score” prediction model
(Altman, 1993). This revised model, Altman felt, significantly improved the predictive ability of his earlier model.
Later, in 1980, Ohlson published a study using "Logit" or Multiple Logistic Regressions in constructing a
bankruptcy prediction model. Ohlson felt that his study had an important advantage in that they had an important
timing advantage that allowed one to check whether the company entered bankruptcy prior to, or after the date of
release for the financials. Ohlson claimed that previous studies did not explicitly consider the timing issue.
According to Cybinski (2001), another difference between the early and late studies was that early studies were
concerned with explanation rather than prediction.

The Journal of Applied Business Research – November/December 2015 Volume 31, Number 6
Copyright by author(s); CC-BY 2071 The Clute Institute
In their study of bankruptcy in Thailand, Pongsatat, et al (2004) found that by the end of the 1970s, most
bankruptcy prediction used some form of multivariate analysis with multiple discriminant analysis the preferred
method. They found also that some researchers felt that multiple discriminant analysis had two fundamental
weaknesses. According to Jones (1987) multiple discriminant analysis does not consider prior probabilities and also
assumes an equal probability of group membership based on sample proportions. To eliminate these two
weaknesses, researchers began to use two additional statistical techniques: multiple logistic regression (or logit
analysis) and probit analysis. Kleinbaum & Klein, (2002) describe logit as a technique based on a cumulative
probability function that does not require the independent variables to be normal. Logit puts a weight on each of the
variables so that the formula generates a probability of classification putting the weighted groups into one or more
separate groups. The probit method is similar to logit except that it uses a nearly identical normal cumulative
probability function instead of the logistical cumulative function (Gentry, Newbold and Whitford, 1985). Boritz and
Kennedy (1995) described the differences between the two approaches as simply being that the logit model used the
cumulative logistic function and probit used the cumulative normal distribution.
In their review of the literature, Pongsatat et al concluded that Martin (1977) probably was the first to
conduct a study using logit analysis in bankruptcy prediction. Martin looked at 58 failed Federal Reserve member
banks and compared them to a group of non-failed banks. Using 25 financial ratios classified as asset risk, liquidity,
capital adequacy and earnings, Martin developed a model that correctly classified failed banks 87% to 96% of the
time. Pongsatat et al concluded that this ability to accurately predict failed financial firms was important in any
analysis of bankruptcy in Thailand.
As stated above, James Ohlson (1980) is acknowledged to be the first researcher to conduct a
comprehensive study of bankruptcy using logit analysis. Ohlson felt that the strength of his technique was that it
was simple to apply and could be used in a number of different circumstances (Ohlson, 1980). Ohlson did
acknowledge that the weakness in his model was that it did not consider the firm’s market transaction data.
Ohlson cited three primary problems with prior studies that had been done using the more popular multiple
discriminant analysis technique. First he objected to the statistical requirements imposed on the distributional
properties of the ratio. Among these requirements were that the variance – covariance relationships of the ratios had
to be the same for both groups and that the predictors (ratios) had to be normally distributed. Second, the output to
the multiple discriminant analysis is a score, which has little intuitive interpretation. Third, he did not feel that the
use of the procedure of matching failed and non – failed firms provided any benefit to an analysis. Ohlson felt that
the use of conditional logit analysis essentially avoids all the problems discussed with respect to multiple
discriminant analysis (Ohlson, 1980). Ohlson conducted three sets of computations using his logit model. Model
one predicted bankruptcy within one year; model two predicted bankruptcy within two years, given that the
company did not fail within the subsequent year; and model three predicted bankruptcy within one or two years.
By obtaining data from the Wall Street Journal Index in the Seventies (1970 – 1976), financial information
of 105 bankrupt firms, and 2058 non-bankrupt firms were obtained from the Compustat file. From Ohlson’s study
the results indicate that the four factors derived from financial statements that are statistically significant for
purposes of assessing the probability of bankruptcy are first, size; second, a measure of leverage (total liabilities to
total assets); third, a measure of performance (net income to total assets and fixed assets to total liabilities); and
fourth, some measures of current liquidity (working capital to total assets, current assets to current liabilities).
(Ohlson, 1980)
By the end of his study Ohlson concluded that the predictive power of the model depends upon when the
financial report is made available, and the predictive powers seem to be robust across the estimation procedure.
Finally, Ohlson recommended that further research needed to be conducted to improve the accuracy of the
prediction model. While Ohlson’s results were not as good as Altman’s, he concluded that his methodology was
more sound. He also reached another interesting conclusion from his study in that he found that the size of the firm
was the most important predictor in his model (Patterson, 2001). According to Khunthong (1997), Ohlson’s model
for predicting bankruptcy within one year, using nine accounting ratios and a cutoff point equally weighted with
type I and type II errors, gave a correct classification 96% of the time.