CORPORATE FINANCIAL DISTRESS AND BANKRUPTCY PREDICTIOIN:

A COMPREHENSIVE ANALYSIS IN PAKISTAN

Muhammad Shahzad Ijaz

11-arid-3600

Supervised by: Mr. Ahmed Imran Hunjra

UIMS-PMAS Arid Agriculture University

Rawalpindi, Pakistan
 OUTLINE
 INTRODUCTION
 LITERATURE REVIEW
 METHODS
 RESULTS
 RECOMMENDATIONS
 LIMITATIONS AND FUTURE RESEARCH

2
 Introduction
 No two subject specialists are in agreement on a single definition of business failure and
financial distress (Bruno and Leidecker, 2001) .
 In the history, different terms have been used by experts to depict the situation of
financial distress.
 Economic failure, Business failure, Technical insolvency, Insolvency in Bankruptcy,
Bankruptcy (Altman and Hotchkiss, 2006)
 These terms are used interchangeably in the literature and also in this study these

terms are used identically.

 Predicting financial health of enterprise at appropriate time permits investor, management,
business counterparts and other stakeholders to be proactive and to formulate efficient
policies.
 All interested stakeholders of the company remain interested in the development of

new prediction models according to new corporate practices to forecast the business
failure.
 So, a model that can correctly forecast business failure in time would be quite useful

for the stakeholders of the company and it would be facilitating companies to

determine their next strategy and future course of action.

3
Use of Financial Ratios in Predicting Financial Distress

 Financial ratio is a quotient of two numbers. These numbers are the items of
financial statements.
 The use of financial data and accounting numbers to analyze performance dates
back to almost 100 years.
 Formation of current ratio was the starting point of ratio analysis with the only
objective of evaluating credit worthiness.
 Probably the earliest study employing accounting data for making decisions about
the credit position of the firm was published in 1908. The main concern was the
extension of bank loans (Rosendale, 1908).
 Previous bankruptcy studies have identified that different financial ratios of
bankrupt companies have the predictive ability to forecast financial distress (Beaver
1966; Altman, 1983; Foster, 1986 and Jones, 1987).

4
Uses of Z-Score in History
Use
Shrieves and Stevens
(1979); Lasfer et al. used Z-Score as a proxy for the risk of bankruptcy in the
(1996); Sudarsanam and field of divestments and merger and acquisitions.
Lai (2001)

Dichev (1998); Ferguson used Z score for the pricing of capital assets and market
& Shockley (2003) efficiency.

used Z score for the determination of appropriate

Wald (1999); Kao (2000) capital structure and distressed securities.

Caouette et al. (1998) used Z score for rating portfolios and bond ratings.

Carcello et al., (1995) and used Z score for assessing the financial health of the
Taffler et al., (2004) companies for going concern research.

5
 Financial distress and bankruptcy prediction models also serve the purpose
of credit evaluation and used as credit scoring models. Credit risk
assessment models have become more crucial due to the implementation of
Basel II.
 These models have proved to be a reliable tool for failure prediction in a
variety of contexts, in different countries. However, such a study is rare in
Pakistan. Thus, it would be a worthwhile to look that how the widely used
models, which were originally developed in developed economy of US, can
be expanded and applicable in less developed / developing country of
Pakistan.

6
 Need of the Study
Early warning models provide the company with a powerful helping Harlan and
tool to identify and manage the supplier firms bankruptcy and prevent Marjorie,
crises in its procurement. (2002)
Although one can expect that the independent auditors or other
Altman and
decision-makers are able to make a correct assessment of the
Saunders,
company’s financial condition. Studies have shown that, in practice,
(1998)
auditors and decision makers do not perform well.

Bankruptcy is a global phenomenon which occurs in developing

Newton,
countries as well as in developed countries also. It cannot be separated
2009; Her and
from doing business. But it occurs at a higher rate in developing
Choe, (1999)
economies as compared to developed ones.

In the year 2012, there were 63 companies which were delisted from
KSE Web
KSE till August.

7
 To protect the investor from loss, a number of studies have been conducted all
over the world and provide various efficient methods for the identification of
failing firms.

 Rashid and Abbas (2011) were the first who studied bankruptcy prediction in
Pakistan using only single model developed through multiple discriminant analysis
and none of the other research has focused Pakistan specifically.
 So, this study can be considered as a step forward or initial step to fulfill the gap
in this area considering two prediction models.

There is a need of investigation that which ratios best predict the

corporate failure. Therefore, this study recognizes the need for the
development of bankruptcy prediction models unique to the Pakistani
environment to keep the corporate stakeholders safe from losses.

8
 Objectives of the Study
The purpose of this study is:

1. To identify the financial and accounting variables which best discriminate

bankrupt companies from non bankrupt companies.
2. To develop Z-Score model through identical procedure of Altman (1968)
using multivariate statistical technique of discriminant analysis.
3. To develop probabilistic bankruptcy prediction model as of Ohlson (1982)
using binary regressand technique of Logit analysis.
4. To make an investigation about the practical applicability of these models for
Pakistani firms according to current accounting and financial practices.
5. To study the efficiency of each model and to compare the efficiency and
prediction results with each other.

9
Significance of the Study
 Lenders This study tends to prove most beneficial in lending function. Lending
institutions and banks use credit scoring and financial distress prediction models to
assess the financial position of their credit customers.
 Accounts Receivable Management: Analogous to banks, corporate account
receivables managers calculates the repayment estimates and bad debts.
 Investors: Investors of the company are interested in the profit prospect of the
company.
 Security Analysts: Debt analyst is, and should be, far more focused on the
downside possible movement of securities. One or more distressed prediction
models would seem to be a careful addition to the security analysts’ process.
 Regulators: It is quite common for a nation’s central bank to utilize its own credit
scoring evaluation systems to assess the credit quality of bank’s portfolios.
 Employees: are interested in their job security and stability of the company.
 Management: of the company are interested in knowing what problems they are
about to face

10
 Literature Review
Author and year Model Predictive Country Firm type
Accuracy %
Altman (1968) MDA 95 USA Manufac. Ind. (ltd.)
Altman et al. (1977) MDA 92.8 USA Manf. &retail (ltd.)
Ohlson (1980) Logit 96 USA General
Aziz et al. (1988) Logit 91.8 USA Mix. Ind. (ltd.)
Back et al. (1996) MDA 85.14 Finland Mix. Ind. (ltd.)
Back et al. (1996) Logit 96.49 Finland Mix. Ind. (ltd.)
Dimitras et al. (1999) MDA 90 Greece Mix. Ind. (ltd.)
Dimitras et al. (1999) Logit 90 Greece Mix. Ind. (ltd.)
Zavgren (1985) Logit 82 USA Mix. Ind. (ltd.)
Ward (1994) Logit 92 USA Non-Fin. Firms
Taffler (1983) MDA 97.8 UK Manufac. Ind. (ltd.)
Deakin (l972) MDA 79.5 Moody Industrial General
Manual
Blum(l974) MDA 94 USA Industrial Firms
Westgaard and Wijst (2001) Logit 97.3 Norway Mix. Ind. (ltd.) 11 11
 Research Model and Ratios of the Study
MDA Model

Z = Discriminant score Wn = Discriminant weight (coefficients)

Xn`= Discriminant variable (ratios)

Logit Model

P(Z) = Probability of bankruptcy Wn = Logit weight (coefficients)

Xn`= Logit variable (ratios)

Ratios of the Study

Thirty seven ratios of this study were classified under six categories i.e.
profitability, liquidity, leverage, asset efficiency, growth and size. The criteria for the
selection of these ratios were their popularity in bankruptcy prediction literature.

12
THEORETICAL FRAMEWORK
Inputs Output

Profitability Ratios

Liquidity Ratios
(Solvency)

Leverage Ratios Specific classifier:

(Debt)

Failure vs. Non-failure

Asset Efficiency
(Management or Turnover)

Growth

Size
Independent variables Dependent variable

13
Hypotheses Statements
 H1: There exists a significant difference in profitability ratios for bankrupt and non
bankrupt enterprises five years prior to bankruptcy.
 H2: There exists a significant difference in liquidity ratios for bankrupt and non
bankrupt enterprises five years prior to bankruptcy.
 H3: There exists a significant difference in leverage ratios for bankrupt and non
bankrupt enterprises five years prior to bankruptcy.
 H4: There exists a significant difference in asset efficiency ratios for bankrupt and
non bankrupt enterprises fives year prior to bankruptcy.
 H5: There exists a significant difference in growth ratios for bankrupt and non
bankrupt enterprises five years prior to bankruptcy.
 H6: There exists a significant difference in size ratios for bankrupt and non
bankrupt enterprises five years prior to bankruptcy.

Additionally, this study also develops Z-Score and logit models for listed companies in
the non financial sector to assess their financial status.
 H0: Z-Score and Logit do not predict bankruptcy in Pakistan.
 HA: Z-Score and Logit do predict bankruptcy on the basis of published financial
information.

14
Materials and Method
This study follows a paired sample design.
Study Techniques

Beaver (1966) Univariate Analysis

Altman (1968) MDA

Ohlson (1980) Logit Analysis

Lennox (1999) and
Logit and Probit Analysis
Menard (1995)
Shumway (2001)
Hazard Model
Abdullah, Halim, Ahmad
MDA, Hazard Model And Logistic Regression
and Rose (2008)
Tan and Kiang (1992),
Neural Networks and Decision Trees
and Jain and Nag (1998)

15
 Sample
There are two groups
Group I: 35 Failed Companies 1
Group II: 35 Non-Failed Companies 0
Selection of Failed Firms
Keeping the topic of the current study and the previous studies on corporate failure
prediction in mind, criteria for ranking a company as a failure were finalized. A
company will be considered failed if it is:
1. Delisted due to winding up under court order / liquidation from KSE.
2. Wound up by the Security and Exchange Commission of Pakistan.

 Complete list of delisted companies was obtained from the website of KSE from
2003 to 2012. A second list of wound up companies by SECP was obtained from
SECP website. Several firms were dropped from the list due to the unavailability
of their data.

16
 Sample
Selection of non-failed firms
Non failed firms were selected based on the following criteria.
1. Company is from the same year as of the failed company in the group I.
2. Company belongs to same sector.
3. Asset size of the company is equal to the asset size of the companies in the group I.
Following procedure was adopted for the non failed enterprises selection
1. Starting from first bankrupt company, year of failing and sector of the failed company was
considered.
2. A matched non failed company was searched on the basis of same sector and year.
3. Within the same sector, a company was selected for group II if the asset size of the company
was almost equal to the asset size of failed firm.
4. If the asset size is not same as to the asset size of the failed company, then select the company
which is closest in asset size.
5. Repeat this procedure until the group II has the thirty five candidates.

This procedure gave birth to the paired sample design This design is in line with the bankruptcy
prediction studies of the history.

17
 Sector-wise Classification of Failed and Non Failed Companies
Each Group
Sector (Failed and Non-Failed) Total
Textile 17 34
Other Sectors 6 12
Fuel and Energy 4 8
Food and Sugar 3 6
Chemical 2 4
Transport Comm & Motor Vehicle 2 4
Engineering 1 2
Total 35 70
Number of Failed and Non Failed Companies
Each Group
Year Total
(Failed and Non-Failed)
2012 8 16
2011 8 16
2010 8 16
2005 3 6
2006 4 8
2003 4 8
18
Total 35 70
 Data
Data of this study for financially failed and non failed companies were extracted
from the State Bank’s balance sheet analysis of joint stock companies listed on
KSE.
5 volumes of Balance sheet analysis were scanned for the data i.e. 2004, 2005,
2009, 2010, 2011

Collection of Financial Statement Data

1 Current Assets 9 Operating Profit
2 Current Liabilities 10 Quick Assets
3 Earnings Per Share 11 Retained Earnings
4 Earnings before Interest and Taxes 12 Sales
5 Fixed Assets 13 Shareholder's Equity
6 Gross Profit 14 Total Assets
7 Long-term Debt 15 Total Liabilities
8 Net Profit

19
 Computation of Ratios

 For all seventy companies, thirty seven ratios were calculated. The criterion for
selecting the ratios was Popularity of that ratio in the bankruptcy prediction
literature.

Category Measures Ratios

EBIT/CL, EBIT/S, EBIT/TA, GP/S, NP/SE, OP/S, OP/TA, RE/TA,

Profitability 12
OP/TL, EPS, EBIT/FE, NP Margin
Liquidity 7 WC/LTD, QR, CA/TA, CA/CL, EBIT/TL, TL/TA, CL/CA,
Leverage 7 CL/TA, SE/LTD, FA/SE, LTD/TD, D/E Ratio, TD/TA, SE/TA
Asset Efficiency 6 ROA, WC/TA,S/CA, S/FA, S/TA, WC/S
Growth 3 Sales G, NP G, TA G
Size 2 Ln TA, Ln TS

20
 Correlation Analysis
 Due to that comprehensive list, there is a likelihood of high degree of correlation
or multicollinearity among the calculated ratios as stated by Altman (1968).
 Threat for successful implementation of discriminant analysis

 Following ten ratios were deleted from the analysis having correlation higher

than 0.8 as Zhang, Chen and Yen (2001).

Group Ratio Description Source

Profitability EBIT/TL Earnings before Interest and Taxes / Total Liabilities Gu

Profitability OP/TL Retained Earnings / Total Liabilities Ohlson

Leverage SE/TA Shareholder's Equity / Total Assets Ugurlu,
Leverage TD/TA Total Debt / Total Assets Eljelly
Liquidity TL/TA Total Liabilities / Total Assets Ohlson
Liquidity WC/TA Working Capital / Total Assets Altman
Profitability EPS Earnings Per Share Madalina
Profitability EBIT/FE Earnings before Interest and Taxes / Total Assets Madalina
Asset Efficiency ROA Return on Assets Ohlson
Profitability NP Margin Net Profit / Sales Eljelly
21
 TECHNIQUE OF THE STUDY
I t-test

 t test was used in SPSS to test the equality of means of remaining twenty seven financial
ratios for two groups of bankrupt and non bankrupt enterprise for five years prior to
bankruptcy.

 It tests equality of means with

H0: μ1 − μ2 = 0
H1: μ1 − μ2 ≠ 0
 If t value is significant at p value <= 0.05 which means that we reject H0 and

assumption of equal mean is violated.

 If the p value is insignificant i.e. greater than 0.05 we can infer that we are failed to

reject H0 and underling mean is homogeneous among bankrupt and non bankrupt
groups.

22
 TECHNIQUE OF THE STUDY
 Aziz and Dar (2006) ranked the sixteen techniques which have been used
heavily in the bankruptcy prediction literature
 MDA was the Top one

 Logit was the Second

Altman (1968) and Ohlson (1980) were the pioneers in using these techniques
and developed Z- Score and O-Score models using MDA and Logit.
II Multiple Discriminant Analysis
 Multivariate statistical methods
 This technique is used to make a forecast in situations where the dependent variable
is in qualitative form, e.g., bankrupt or non-bankrupt, male or female.
 The purpose of discriminant analysis is to find the linear combination of ratios
which best discriminates between the groups.
 The individual variables are transformed into the single discriminant score which is
also called z-score and then it used to classify the objects.
 This classification is dependent upon the characteristics of the groups.
 SPSS V20 was used for this analysis.
23
 Stepwise procedure was adopted.
 List of ratios analyzed in discriminant analysis along with sources
Ratios Description Source
EBIT/CL Earnings before Interest and Taxes / Current Liabilities Gu (2002)
EBIT/S Earnings before Interest and Taxes / Sales Ugurlu (2006)

Profitability
EBIT/TA Earnings before Interest and Taxes / Total Assets Altman (1968)
GP/S Gross Profit / Sales Ugurlu (2006)
NP/SE Net Profit / Shareholder's Equity Ugurlu (2006)
OP/S Operating Profit / Sales Eljelly et al. (2001)
OP/TA Operating Profit / Total Assets Eljelly et al. (2001)
RE/TA Retained Earnings / Total Assets Altman (1968)
WC/LTD Working Capital / Long-term Debt Ugurlu (2006)
Liquidity

QR Quick Assets / Current Liabilities Zmijewsky (1984)

CA/TA Current Assets / Total Assets Ugurlu (2006)
CL/CA Current Liabilities / Current Assets Ohlson (1980)
CA/CL Current Assets / Current Liabilities Zmijewsky (1984)
CL/TA Current Liabilities / Total Assets Zmijewsky (1984)
Leverage

SE/LTD Shareholder's Equity / Long-term Debt Ugurlu (2006)

FA/SE Fixed Assets / Shareholder's Equity Eljelly et al. (2001)
LTD/TD Long-term Debt / Total Debt Eljelly et al. (2001)
D/E Ratio Total Debt / Shareholder's Equity Zeitun et al. (2007)
S/CA Sales / Current Assets Ugurlu (2006)
Efficiency
Assets

S/FA Sales / Fixed Assets Eljelly et al. (2001)

S/TA Sales / Total Assets Altman (1968)
WC/S Working Capital / Sales Eljelly et al. (2001)
Sale G Sales Growth (St - St-1) / St Madalina (2009)
Growth

NP G Net Profit Growth (NPt - NPt-1) / NPt Madalina (2009)

TA G Total Assets Growth (TAt - TAt-1) / TAt Madalina (2009)
LnTA log (Total Assets) Ohlson (1980)
Size

LnTS log (Total Sales) Madalina (2009) 24

 TECHNIQUE OF THE STUDY
III Logit model

 Shumway (2001) stated that logit model calculates the probability of the firm
to join bankrupt or non bankrupt group with k predictors x1, x2, x3 … xk
which are financial and accounting ratios.
 One crucial issue in binary dependent variable models is the determination
of cutoff point.
 When binary variable consists of two groups e.g. non bankrupt or bankrupt,
0.5 probability is frequently used in the literature as cutoff point for
classification accuracy as Wang and Campbell (2010) and Madalina (2009).
 Forward looking procedure was adopted to identify the best predictor
financial ratios for the Logit model Madalina (2009).
 Eviews 6 was used for this analysis.
 Only variables which pass the mean dissimilarity test were
selected as the input for the Logit model.

25
 Results: Significant Differences

Years Prior to Bankruptcy

5/12 Profitability Ratios
5 4 3 2 1

Earnings before Interest and Taxes / Current t value -.225 -.166 -.566 1.790 -1.343
Liabilities p value .022 .202 .015 .313 .134

t value -.084 -1.046 -.294 -.989 -1.807

Earnings before Interest and Taxes / Sales
p value .138 .314 .007 .066 .072

Earnings before Interest and Taxes / Total t value -.239 -.259 -.640 -.027 -.170
Assets p value .031 .051 .069 .823 .178

t value -.041 -1.122 -.173 -1.009 -.553

Gross Profit / Sales
p value .426 .224 .025 .227 .049

t value -.259 -.199 .007 .066 -.069

Operating Profit / Total Assets
p value .036 .119 .982 .607 .554

H1Accepted 26
 Results: Significant Differences
Years Prior to Bankruptcy
1/7 Liquidity Ratios
5 4 3 2 1

t value -96.618 -90.277 -98.110 -70.708 -84.109

Quick Assets / Current Liabilities
p value
.000 .000 .000 .000 .000

Years Prior to Bankruptcy

2/7 Leverage Ratios
5 4 3 2 1
t value .012 .045 -.068 -.059 -.079
Long-term Debt / Total Debt
p value .831 .453 .204 .295 .058
t value -146.34 -51.87 -39.99 -65.81 -50.66
Total Debt / Shareholder's Equity
p value 0.00 0.24 0.37 0.01 0.11

Years Prior to Bankruptcy

1/6 Assets Efficiency Ratios
5 4 3 2 1
t value -96.62 -90.28 -98.11 -70.71 -84.11
Sales / Current Assets
p value .00 .00 .00 .00 .00

Years Prior to Bankruptcy

1/2 Size Ratios
5 4 3 2 1
t value -1.373 -2.158 -2.040 -2.561 -2.700
log (Total Sales)
p value .016 .003 .002 .000 .000

H2, H3, H4, H6 Accepted 27

 Results: MDA

The output from the program consisted of a set of discriminant weight

which indicates the linear combination of variables which maximizes
the differences between the groups as well as a vector which indicates
the relative contribution of each variable.

Z-ScorePakistan = 0.557 X1 - 0.420 X2 + 0.477 X3 + 0.895 X4 + 0.638 X5

Where,
X1 = EBIT / Sales X2 = Long-term Debt to Total Debt
X3 = Sales to Current Assets X4 = Log of Total Sales
X5 = Quick Ratio

28
 Results: MDA
Cut off point for classification accuracy = 0

Goodness of fit statistics for new model

Test of Function(s) Wilks' Lambda Sig.

1 .402 .000

Classification accuracy

Predicted Group Membership

B/NB Total
.00 1.00
.00 35 0 35
Count
1.00 4 31 35
Original
.00 100 0.0 100.0
%
1.00 11.4 88.6 100.0
94.3% of original grouped cases correctly classified.

29
 Results: Logit Analysis

Where, X1 = Log of Total Sales X2 = Long-term Debt to Total Debt ratio

X3 = Operating Profit to Total Asset ratio X4 = Sales to Current Assets ratio and X5 = Quick ratio
Variables C LNTS LTDTD OPTA SCA QR
Coefficient -40.49 1.93 -0.46 10.61 9.9 0.25
Prob. 0.01 0.01 0.02 0.02 0.02 0.02

Goodness of Fit Statistics for New Developed Model

Count R2 = 95.71% LR stat = 80.8
Mc Fadden R2 =83% p value < 0.000
Classification Accuracy
Estimated Equation
Dep = 0 Dep = 1 Total
P(Dep=1)<=C 34 2 36
P(Dep=1)>C 1 33 34
Total 35 35 70
Correct 34 33 67
% Correct 97.14 94.29 95.71
30
% Incorrect 2.86 5.71 4.29
Predictive accuracy of MDA and Logit models

Correct Predictions Incorrect Predictions

count % count %

Non Bankrupt 35 100 0 0

Z-Score
MDA /

Bankrupt 31 88.6 4 11.4

Overall 66 94.3 4 5.7

Non Bankrupt 34 97.14 1 2.86

Logit

Bankrupt 33 94.29 2 5.71

Overall 67 95.71 3 4.29

31
 Conclusion

Support
Ugurlu and Aksoy (2006)
Profitability Ratios GP/S, EBIT/Sales, EBIT/CL,
Altman (1968)
OP/TA, EBIT/TA
Ohlson (1980)
Liquidity Ratios QA/CL Edmister (1972)
Leverage Ratios LTD/TD, TD/SE Ugurlu and Aksoy (2006)
Asset Efficiency Ratios S/CA
Size Ratios log (Total Sales) Madalina (2009)
Yang et al. (1999); Kidane
Z-Score Model 94.30 percent (2004); Sori and Jalil (2009)
Sung et al. (1999)

Westgaard and Wijst (2001);

Logit Model 95.71 percent
Abdullah and Halim (2008)

32
Recommendations
 Credit Evaluation
 Enterprise Internal Control
 Creditors, Fund Managers and All Shareholders
 Government and Market Authorities
Limitations of the Study
 Qualitative factors as related to management, labor, market
conditions and other macroeconomic conditions that have an impact
on the performance of enterprises.
Suggestions for Future Research
 Testing the models out of the sample accuracy.
 Extended to hazard model to incorporate time varying effects of the
difference in time when the company goes bankrupt.
 Inclusion of macroeconomic variables.

33
Thank You !
 Discussion on Model Ratios

Model Ratios and Signs Support

EBIT / Sales (+) Theodossiou (1996)
Long-term Debt to Total Debt (-) Ugurlu and Aksoy (2006)
Sales to Current Assets (+) Etemadi (2008)
Log of Total Sales (+) Madalina (2009)
Quick Ratio (+) Edmister (1972)

Operating Profit to Total Asset

Fufa (2011)
ratio (+)

35
 Operating profit is the EBT

EBIT / Sales (+) the ability of a firm in covering all production costs and providing
some reasonable margin of profit.
Long-term Debt to Total Debt (-) heavy reliance on short-term debt in failed firms
indicate severe liquidity problems that increase the risk of default
Sales to Current Assets (+) current asset efficiency. This ratio measured the
utilization of asset to generate sales of the company.. A higher efficiency means
higher profitability and better liquidity and finally to lower default risk.
Log of Total Sales (+) cost coverage and increases profitability
Quick Ratio (+) it the short term liquidity

Operating Profit to Total Asset ratio (+) This ratio showed the earning power of
the assets which was an important factor for the existence of the company and in
predicting bankruptcy.
Table 1: Variables in the Analysis
Sig. of Table 2: Cut-off point for Z-score model
Toler Wilks'
Step ance
F to
Lambda Function 1
Remove
Financial Status Z-Score
1 Log (Total Sales) 1.00 .000 Bankrupt -1.203
2 Log (Total Sales) .996 .000 .779 Non-Bankrupt 1.203
Quick Ratio .996 .000 .664
Table 3: Variables Entered / Removed a, b, c
3 Log (Total Sales) .945 .000 .713
Exact F
Quick Ratio .993 .000 .571 Variables Lambd
EBIT/S Step
.947 .002 .543 Entered
4 Log (Total Sales) a Statistic Sig.
.944 .000 .620
Quick Ratio .896 .000 .556 Log (Total Sales) .664 34.334 .000
1
EBIT / Sales .947 .004 .493
2 Quick Ratio .543 28.159 .000
Long-term Debt/
.900 .021 .472
Total Debt
3 EBIT / Sales .472 24.649 .000
5 Log (Total Sales) .734 .000 .620
4 Long-term Debt/
Quick Ratio .893 .000 .513 Total Debt
.434 21.194 .000
EBIT / Sales .862 .001 .478 5 Sales / Current
.402 19.054 .000
Long-term Debt/ Assets

Total Debt .831 .006 .453 At each step, the variable that minimizes the overall Wilks'
Lambda is entered.
Sales / Current a. Maximum number of steps is 56.
Assets .701 .027 .434 b. Maximum significance of F to enter is .05.
c. F level, tolerance, or VIN insufficient for further computation.
Normality Test: A basic assumption of MDA is the normality of the classifying
variables (Kleinbaum, Kupper, & Muller, 1988) but it is applicable just when the
sample size is less than 50 observations (Gu, 2002). Since in this study the sample
size was 70(firms) then the normality test was not necessary.

Type I error (the probability that a firm predicted not to fail will fail) and Type II error (the
probability that a firm predicted to fail will not fail
1= non-failed co’s predicted to fail / total co’s predicted to fail = 0/35 = 0 %
2= failed co’s predicted not to fail / total co’s predicted to fail = 4/35= 11.4 %

Leverage ratios measure the capability of a firm in paying its debt obligations.
 T test: equal variance

 This test assumes that the underlying variances are homogeneous. In SPSS, its output is
shown with both equal variances assumed and equal variances not assumed. If F value
is significant at p value <= 0.05, we used equal variances not assumed. On the contrary,
if p value is insignificant i.e. greater than 0.05, we used first row of output which is
equal variances assumed.
 Logit coefficient Interpretation
Brooks (2008) stated that mean values for each explanatory variable in Logit model should
be calculated for the correct interpretation of coefficients. Therefore, mean values X̅1, X̅2, X̅3,
X̅4 and X̅5 are 16.10, 3.09, 0.20, 0.18 and 28.23 respectively. Now the logistic function was
calculated.

Where, X1 = Log of Total Sales X2 = Long-term Debt to Total Debt ratio

X3 = Operating Profit to Total Asset ratio X4 = Sales to Current Assets ratio and X5 = Quick ratio

So, with one unit increase in LNTS, probability of the case to join group 1, e.g. Yi=1,
will decrease by (-1.93 × 0.47) = 0.91. Thus a one unit increase in LTDTD will cause
a increase in the probability that the outcome corresponding to yi = 1 will occur by
0.46× 0.47 = 0.22. The corresponding decrease in probability to join bankrupt group
for OPTA, SCA and QR are -10.61× 0.47 = 4.98, -9.90× 0.47 = 4.65 and -0.25×
0.47 = 0.12, respectively. These are also called marginal effects and unfortunately, E-
views does not calculate marginal effects. These marginal effects were calculated in a
spreadsheet as suggested by Brooks (2008).
Economic failure: is a situation in which realized rate of return
including the allowances for risk consideration on capital invested by
company is considerably and continuously less than the prevailing rate
of return on comparable investment.
Business failure: According to D&B, this term incorporates cession
of business, foreclosure, and includes voluntarily withdraw from
business with unpaid promises.
Technical insolvency: A situation in which a company is unable to
fulfill its current obligations due to liquidity.
Insolvency in Bankruptcy: It occurs when fair valuation of the assets
of the company fall short than total liabilities. Therefore, the actual net
worth of the company is negative. It is difficult to detect than technical
insolvency as it requires complete valuation analysis.
Bankruptcy: Bankruptcy itself is a formal declaration of bankruptcy
by court as a result of a petition of bankruptcy reorganization or
liquidation of assets.