optiuneva+reorganizare+falim_australia

8/12/2019 OptiuneVA+reorganizare+falim_australia

1/27


2/27

Pacific Accounting Review Vol. 16, No. 1, June 2004 31

An Exploratory Study of the Company

Reorganisation Decision in Voluntary Administration

JAMESROUTLEDGE*DAVIDGADENNE**

A primary purpose of the voluntary administration legislation is to

provide a flexible procedure by which a company can attempt to

reorganise its affairs and continue trading. Informed decision-making

regarding which companies should attempt reorganisation is critical

to the efficient operation of company rescue legislation. This paperexplores decision-making associated with the voluntary administration

process, with a focus on the relevance of financial information to the

reorganisation decision. Statistical models are developed to provide

some insight into the reorganisation decision and the problem of

identifying suitable (successful) reorganisation candidates from a pool

of distressed companies. Additionally, insolvency experts decisions

regarding companies prospects in reorganisation are examined. The

decision accuracy of insolvency experts was found to be significantly

lower than statistical model accuracy, indicating that further

development of statistical models may be a useful aid to insolvency

experts.

* James Routledge is at the School of Business, Bond University

** David Gadenne is at the Faculty of Business and Law, Central QueenslandUniversity

The authors would like to thank delegates from the 2002 AAANZ conference in

Perth, two anonymous reviewers and Martin Lally (the editor) for their helpful com-ments and constructive advice.


3/27

32 An Exploratory Study of the Company Reorganisation Decision in Voluntary Administration

(1) INTRODUCTION

The voluntary administration (VA) insolvency procedure1was introduced as part ofAustralias corporate insolvency regime in June 1993. The objectives set out for the

VA regime were twofold. First, VA was intended to provide a flexible and relativelyinexpensive procedure by which an insolvent company could attempt to formulate anarrangement with creditors to rescue the companys business. Second, when rescue isnot possible, the VA procedure was intended to provide a flexible winding-up processso as to achieve a greater return to creditors than would result from immediateliquidation.2The Australian approach to company rescue legislation was heavilyinfluenced by the existing restructuring regimes operating in the United Kingdomand United States.3

A study by Hodson and McEvoy (1995) estimated that 20 percent of companies thatenter VA attempted to reorganise their affairs and continue trading. Despite theextensive use by distressed companies of the reorganisation provision in VA, there isa lack of empirical research that has examined its operation. Accordingly, an importantcontribution of this paper is that it provides one of the few references on thereorganisation alternative available under VA. With any company rescue orrehabilitation legislation the decision as to which companies should attemptreorganisation is critical to its efficient operation. This is because insolvency legislation,which provides the opportunity for company reorganisation, can create adverseincentives to prolong the existence of non-viable firms (Martel, 1991). Our analysisfocuses on how the information content of historical financial information can explainand inform decision-making associated with the VA procedure. The motivation forthis research is to provide a reference that will serve to inform future decision-makingassociated with the VA regime.

(2) METHOD

Exploring the reorganisation option for companies that enter VA requires considerationof two events. The first event, which we have called the reorganisation event, occurswhen the decision is made as to whether a company that has entered VA will attemptto reorganise or proceed to liquidation. The second event, which we have called the

1Part 5.3A of the Australian The Corporations Law (Cth).

2The objectives are outlined in s 435A of The Corporations Law (Cth).

3

In the United Kingdom, the Insolvency Act 1986 (UK), and in the United States Chapter 11 of theBankruptcy Code 1978 (US). For comparative discussion of the regimes, refer to Crutchfield (1994).


4/27

Routledge and Gadenne 33

Diagram 1 Summary of Events in Voluntary Administration

performance event, is the event of success or failure for companies that proceedwith reorganisation. The two events and their relationship to the VA procedure aresummarised in Diagram 1 below.

In this paper, we explore decision-making at the time of the reorganisation event by

conducting multivariate statistical analysis. Using a sample of companies that haveentered VA, we perform a logistic regression to distinguish companies that reorganisefrom companies that liquidate. The purpose of this analysis is to identify the financialcharacteristics, if any, that are relevant to determining whether a company will proceedwith reorganisation upon entering VA.

Statistical analysis of the performance event is conducted that is intended to distinguishbetween successful and unsuccessful companies that attempt to reorganise. For thisanalysis we review the post VA financial performance for each company that attemptsreorganisation and classify its performance as either successful or unsuccessful.Logistic regression analysis is then conducted to identify financial characteristicsthat distinguish between the successful and unsuccessful reorganised companies. Thisanalysis is informative in identifying the unique financial characteristics, if any, ofsuitable reorganisation candidates from a pool of distressed companies.

For the performance event we also conduct an experimental task that compares theperformance of the statistical analysis with that of human decision-makers in

distinguishing companies that reorganise successfully. Jointly examining financialdistress decisions using statistical models and human information processing is well


5/27


Houghton and Woodliff (1987, pp.538-539) noted that a critical aspect of the LensModel is that imperfect human decisions have two causes. First, they may be a productof imperfect information describing the criterion variable, that is, the informationcues fail to reflect exactly the criterion event. Second, they may be attributed to sub-optimal processing of data or imperfect cue utilisation by the decision-maker.Comparing the performance of statistical analysis and human judgement in identifyingsuccessful reorganisation candidates will provide some insight into the relevance ofstatistical models as an aid to decision-makers in VA.5

Diagram 2 The Lens Model Framework

Information Cues

(Accounting Ratios)

Environmental

Predictability

Model/Human

Utilisation

4Libby (1975) provided one of the first examples of the application of the Brunswik Lens Model frameworkin accounting research. He analysed the judgement accuracy of bank loan officers in a business failure

prediction task. Subsequent studies have extended the application of Libbys seminal work (see, for example,Zimmer 1980, Abdel-khalik and El-Sheshai 1980, Casey 1980 and 1983, and Houghton 1984). Otherstudies have jointly considered the performance of statistical models and neural network expert systems in

bankruptcy prediction tasks. See, for example, Lenard et al. (1995).

5Successful application of prediction models have been reported in similar studies concerned with auditor

assessments of going-concern status for companies (see, for example, Altman and McGough 1974, Deakin1977, Kida 1980, Simnett and Trotman 1989, and Koh 1991).

established,4and has generally been based on the Brunswik Lens Model (see Brunswik1952). The Lens model approach is used in this study and its application is summarisedin Diagram 2 below.


6/27


Determining suitable predictor variables to include in this type of study is problematicdue to the lack of a theoretical basis on which selection can be based. The selection ofvariables for the development of statistical models and the administration of anexperimental task is therefore subjectively based on a review of prior studies. The

next section reviews prior financial distress studies that are informative with respectto locating a suitable set of predictor variables.

(3) VARIABLE SELECTION

There is a considerable body of prior financial distress research that has focused oninsolvency (or bankruptcy) prediction. These studies have developed and refinedmultivariate financial distress prediction models since the seminal research in this

area by Altman (1968). Prior studies of both Australian listed companies (Lincoln1984, Izan 1984, Castagna and Matolcsy 1981), and unlisted companies (Cybinski1995, Shailer 1990, McNamara et al. 1988) have confirmed the usefulness of financialratio data in the bankruptcy prediction task in the Australian setting. More recently,financial distress studies have focused on the problem of distinguishing between typesof distressed firms rather than distinguishing distressed firms from healthy firms.This research has demonstrated that different types of financial distress arecharacterised by different underlying constructs (Gilbert et al. 1990, Ward and Foster1997). This body of research is particularly informative in respect to variable selectionfor this study.

Frost-Drury et al. (2000) conducted one of the few Australian studies that have focusedon different types of distress in VA. Their study investigated the propositions that (1)companies entering VA could be distinguished from healthy companies and, (2)companies entering VA can be distinguished from companies that directly enteredliquidation. The propositions were tested by logistic regression analysis using financial

predictor variables. Their model results indicated that poor profit performance and

greater proportions of assets tied to working capital distinguished financially distressedfrom healthy companies. Importantly, Frost-Drury reported that the financialcharacteristics of VAs differed from liquidations. Their results indicated that a firmwith greater proportions of assets tied to working capital was more likely to liquidatethan enter VA.

Further research that has focused on distinguishing types of financial distress and itssubsequent resolution has considered the operation of the United States Chapter 11

bankruptcy procedure. Comerford (1976) examined the financial characteristics of

52 firms that had filed bankruptcy petitions under the United States Chapter 11procedure. The objective of Comerfords study was to identify financial characteristics


7/27


that distinguished firms that liquidated from those that successfully reorganised. Hissample comprised 26 firms that had filed Chapter 11 petitions and were subsequentlyliquidated, and 26 firms that had reorganised their affairs and continued operation fora period of two years. Comerfords study used the principal components data reduction

method to extract factors representing financial dimensions for companies in thesample. This approach to determining relevant financial ratio independent variableshas been used extensively since the seminal work of Pinches et al. (1973).6Eighteenoriginal ratios were reduced to six significant factors. The highest loading ratio foreach factor was included as an independent variable in a multivariate discriminantanalysis. The six ratios that comprised the significant discriminant function includedthree ratios representing measures of liquidity, two ratios representing profitabilityand one leverage ratio.

Casey et al. (1986) used financial variables to discriminate between a group ofliquidated firms and a group of restructured firms under the US Chapter 11 bankruptcy

procedure. The variables used in the study were selected based on a model of creditorand equity holder coalition behaviour that had been developed by White (1983 and1989). Casey et al. (1986) found that prior profitability and the percentage ofuncollateralised (free) assets were important predictors of whether a firm wouldliquidate or reorganise.

Campbell (1996) undertook a similar study using a sample of distressed US closely-held firms that entered the Chapter 11 procedure. His study extended the Casey et al(1986) study by using both financial and non-financial variables as predictor variables.Campbell (1996) reported profitability, the percentage of uncollateralised (free) assets,and firm size (measured by assets) to be significant financial predictor variables.With respect to non-financial variables, the study reported that the composition of afirms creditors affected the financial distress resolution. Consistent with earlierfindings by Hotchkiss (1995) and Lo-Pucki (1983), Campbell (1996) found industryclassification to be a significant predictor of the Chapter 11 outcome.

Kennedy and Shaw (1991) also extended the model tested by Casey et al. (1986) byincluding the firms going-concern audit opinion as a predictor of distress resolution.They found limited evidence that distressed firms with unmodified going-concernaudit opinions were more likely to be reorganised than to be liquidated. Kennedy andShaw (1991) reported that the audit opinion had incremental predictive value inaddition to past profitability and free assets. However, the predictive ability of themodel was limited to instances where the firm had filed for bankruptcy within one

6See also Zavgren (1985) for use of this technique.


8/27


year of the issuance of the financial report from which data was taken. Casterella etal. (2000) also applied the Casey et al. (1986) model in the development of a proxyfor auditors expectation of bankruptcy resolution.

The effect of non-financial variables on distress resolution has been considered insome detail. For example, a recent study by Barniv et al. (2002) used financial andnon-financial data to predict whether a firm entering the US Chapter 11 procedurewould be acquired, emerge, or proceed to liquidation. Several non-financial variableswere found to be significant predictors of the Chapter 11 outcome. The study reportedthat size (measured by assets), composition of firm debt, management change, evidenceof fraudulent activity, and change in the market price of securities in the period leadingto the bankruptcy filing, were significant predictors. Prior research has also providedevidence that macroeconomic conditions affect distressed firms and the process of

reorganisation, and that the predictive power of multivariate models can be increasedby the inclusion of macroeconomic variables. For, example Rose et al. (1982) founda strong association between macroeconomic variables and United States bankruptcyrates. Altman (1971) and Levy and Barniv (1987) found that bankruptcy rates in theUnited States correlated with changes in the Gross National Product and general

price levels. Levy (1992) demonstrated analytically the effect of interactions amongfinancial, industrial and macroeconomic factors on the timing of liquidation fordistressed firms.

It is evident from the review of prior studies that numerous financial and non-financialvariables have been found to be useful predictors of financial distress resolution. Inthis study, it has been necessary to limit the number of variables included in ouranalysis for three reasons. First, we are interested in considering the predictive valueof data that can be easily accessed in a companys financial statements. Secondly,extensive use of predictor variables could have a detrimental effect in the experimentaltask due to problems associated with high information load. Thirdly, the relativelysmall sample size would not support large numbers of predictor variables in the

statistical models, and could lead to overfitting.

While there is variation in the financial variables used in prior relevant studies, thesignificant predictors have consistently operationalised the financial constructs ofequity, leverage, liquidity, profitability and size. Accordingly, financial variables foreach of these constructs will be included in the statistical analyses and as informationcues in the experimental task. In addition, we have included industry classification asa non-financial component in the model due to its significance in prior studies.


9/27


(4) STATISTICAL ANALYSES

The independent variables included in the statistical analyses are summarised in Table1.

Table 1Independent Variables

Financial Construct Ratio or Variable

Financial Leverage/ Total Liabilities/Total Assets (TLTA)Equity Commitment Total Liabilities/Owners Equity (TLOE; coded 1 to 5,

1 lowest ratio, 5= negative equity)*

Short-Term Liquidity Current Assets/Current Liabilities (CACL)Current Liabilities/ Total Assets (CLTA)

Earnings Prospects Operating Profit / Total Assets (OPTA)

Industry Classification Indicator Based on ANZSIC Grouping

Size Total Assets (LNTA; natural log of total assets)

* TLTA and TLOE are similar variables, however, both were included as many companies in the sample

had negative owners equity and inclusion of TLOE in the categorical form provides incremental information

regarding the existence and extent of equity commitment

Calculation of the ratio TLOE is problematic because some of the companies havenegative equity values and their ratios would be misleading. To overcome this problem,a categorised form of the ratio was calculated. The categorised variable was codedwith a value of one to five: five representing ratios with a negative equity position,and values of one to four being quartile groups for ratios with a positive equity value.The effect of applying this method of addressing problems with ratio calculation wastested by Cybinski (1995), who found the original ratio information value was

preserved in models developed using categorised independent variables.7 Thedependent variables for statistical models required classification of subject companies

based on (1) whether the company reorganised or was liquidated (for the decisionevent), and (2) whether a company was successful or unsuccessful in reorganisation(for the performance event). For the liquidation/reorganisation dichotomous variable,a company was classified as a liquidation if the administration ended in liquidation,or the administration ended with a deed of company arrangement that provided for

7Cybinski (1995) tested groupings consisting of three and five categories, and found that the five categorygrouping gave a slightly superior goodness of fit of the model to data. Cybinskis study did not aim tofind the best categorisation regime, however, it did indicate that reliable models could be developed usingdiscrete categorisation of financial ratio independent variables.


10/27


winding-up (often referred to as a liquidating deed). The dependent variable wascodedzero for liquidation and onefor reorganisation. The decision event model isshown below.

DECISION = b0 + b1TLTA+b2TLOE+b3CACL+b4OPTA (1)+b

5LNTA

+b

6INDUSTRY + error

For the performance event analysis, companies were classified as successful if theiraverage return on assets over the three years subsequent to reorganisation equalled orexceeded the industry average. Return on assets (OPTA) is a continuous variable;however, we have assigned it a dichotomous value to enable comparison betweenstatistical models and the experimental task results. The dependent variable was coded

zero for an unsuccessful reorganisation and onefor a successful reorganisation. The

performance event model is shown below.

PERFORMANCE = b0+ b

1TLTA

+b

2TLOE

+b

3CACL

+b

4OPTA (2)

+b5LNTA

+b

6INDUSTRY + error

The sample used for the development of models included companies that enteredvoluntary administration between July 1993 and 1995. The selection of this earliertime period was necessary for sufficient post VA performance data to be obtained.Financial data was obtained from company annual returns and financial reports lodgedwith the Australian Securities and Investments Commission. Ratios were calculatedfor the financial year that ended immediately before the companys entry to VA.

Data screening for outliers among independent variables was performed by reviewingthe Studentized residuals and Cooks Distance. One case was identified as an outlierusing these diagnostic techniques. This company had been capitalising large amountsof expenditure up to the time of appointment of the administrator. Considering theunusual nature of the companys financial profile it was deemed appropriate to remove

the case from the sample.8

The final sample of 66 companies was comprised of 33reorganised companies and 33 liquidated companies, and included 45 proprietarycompanies, 17 unlisted public companies and 4 listed public companies.

Correlations among independent variables are reported in Table 2. The highestcorrelation was between total liabilities/total assets and current liabilities/total assets

8The company capitalised $6.4 million of Exploration and Evaluation Expenditure over two years; thisamount was written down to $1.15 million in the year of administration by revaluation of assets and losson disposal of tenements.


11/27


12/27


Results of the reorganisation event analysis are summarised in Table 4. The Chi-Square to remove statistic was used to determine the significance of coefficients inthe model. This test statistic is determined by calculating the amount that the modelfit changes if a variable is removed. Coefficients for the leverage variables (TLTAand TLOE) were both significant atp


13/27


The results may be explained by creditor bargaining and coalition behaviour amongstakeholders affecting the reorganisation decision. This would be consistent with thefinding of coalition behaviour in studies that have considered reorganisation decisionsunder the United States Chapter 11 procedure. Where the levels of debt to equity arehigh, yet some value in equity still exists, management and equity holders would be

likely to form a coalition in support of reorganisation as both parties would be worseoff in liquidation due to the high debt to equity ratio and costs associated with company

Table 4

Logistic Regression Results - Liquidation and Reorganisation Outcomes

Estimated Coefficients for Equation (1)

(Dependent Variable is 0 for Liquidation and 1 for Reorganisation)

Variable S.E. 2 Sig.(n=66) to remove

TLTA 3.658 1.490 10.845 0.001**

TLOE -1.935 0.636 14.538 0.001**

CACL 0.193 0.192 4.339 0.037**

CLTA 5.388 1.839 14.963 0.001**

OPTA -1.465 1.519 0.937 0.333

LNTA 0.548 0.304 3.815 0.051

INDUSTRY 41.744 0.001**

Mining -42.204 36.216

Manufacturing -2.915 1.086

Wholesale 0.597 2.082

Retail -1.524 1.095

Construction 109.60 56.112

Service 0.000

Model Results

-2 Log Likelihood 46.853

2(11, n=66) 44.913Significance 0.000

Nagelkerke Pseudo R2 0.658

** denotes significant atp


14/27


liquidation. Moreover, higher levels of liquidity would provide the company withgreater capacity to bargain with unsecured creditors at the time of the reorganisationevent.

Classification results for the model are presented in Table 5. Overall, the modelcorrectly classified 80 percent of companies in the sample.

Table 5

Classification Table for Liquidation and Reorganisation Outcomes

Predicted Percent Correct

Observed 0 (liquidation) 1 (reorganisation)

0 (liquidation) 26 7 78.79%

1 (reorganisation) 6 27 81.82%

80.30% Overall

(The cut value is 0.50)

Table 6

Model Validation Testing:

Classification Table for Liquidation and Reorganisation Outcomes


Observed 0 (liquidation) 1 (reorganisation)0 (liquidation) 21 12 63.63%

1 (reorganisation) 9 24 72.72%

68.18% Overall


Lachenbruchs (1975) leave one out holdout procedure was employed to test thevalidity of the models predictive ability. Results of the holdout testing procedure are

presented in Table 6. Classification results for the holdout test were lower than for theestimation sample at 68 percent, which may suggest some model overfitting.

9The holdout procedure involves removing each case from the sample, calculating estimation modelcoefficients from the remaining cases and subsequently classifying the holdout case based on the calculatedcoefficients. If overfitting is a problem the prediction model will achieve a substantially higher level of

predictive accuracy on the estimation sample than it does for the holdout procedure.


15/27


4.2 Results - Performance Event Analysis

Summary statistics presented in Table 7 indicate some differences between companiesgrouped according to successful and unsuccessful reorganisation. A t-test was used to

identify significant differences between mean values. Probability values indicate thatmean company profitability (indicated by return on assets) is significantly different

between groups at p


16/27


Table 9

Classification Results For Successful and Unsuccessful Reorganisations


Observed 0 (unsuccessful) 1 (successful)

0 (unsuccessful) 17 2 89.47%

1 (successful) 2 11 84.62%

87.50% Overall(The cut value is 0.50)

Table 8

Logistic Regression Results For Successful and Unsuccessful Reorganisations

Estimated Coefficients for Equation (2)

(Dependent Variable is 0 for Unsuccessful Reorganisation

and 1 for Successful Reorganisation)

Variable S.E. 2 Sig.(n=66) to remove

TLTA 13.673 18.622 3.477 0.062

TLOE 6.607 7.738 4.256 0.039**

CACL 14.667 17.920 13.547 0.001**

CLTA -0.488 3.388 0.022 0.882

OPTA 40.549 55.653 9.662 0.001**LNTA -2.631 4.002 2.720 0.099

INDUSTRY 14.215 0.007**

Mining 17.031 737.205

Manufacturing -1.889 3.182

Wholesale 5.812 736.207

Retail -33.542 218.682

Service 0.000

Model Results

-2 Log Likelihood 11.669

c2(10, n=32) 31.561

Significance 0.0005

Nagelkerke Pseudo R2 0.846

** denotes significant atp


17/27


Results of the Lachenbruch (1975) holdout testing procedure are presented in Table10. The classification results for the holdout test were similar to results for theestimation sample at 83.2 percent.

Table 10Model Validation Testing: Classification Table for Successful

and Unsuccessful Reorganisations


Observed 0 (unsuccessful) 1 (successful)

0 (unsuccessful) 17 2 89.47%

1 (successful) 3 10 76.92%

83.20% Overall


Analysis of the performance event indicates that there is a relationship between theselected variables and the performance event. Therefore, the variables should providerelevant information to decision-makers regarding the prospects of companies thatreorganise under voluntary administration. Comparison of the reorganisation and

performance event analyses suggests that VA decision-making may not be based on a

full consideration of a companys future prospects. Perhaps most striking is the absenceof the past profitability as a significant characteristic of companies for which thedecision is to attempt reorganisation.

The following section of this paper investigates the ability of insolvency experts toutilise the variables (information cues) in a reorganisation decision experimental task.

(5) EXPERIMENTAL TASK

Subjects for the experimental task were Australian insolvency practitioners. Thesepractitioners are an important party to the voluntary administration decision processin their role as appointed administrators. The Corporations Act requires that theappointed administrator assess the prospects of a company and make arecommendation to creditors regarding the companys future.10From a listing published

by the Insolvency Practitioners Association of Australia, contact details forapproximately 136 insolvency firms were obtained. Firms were contacted and

10See section 438A of The Corporations Act 2001(Cth).


18/27


requested to participate in the experimental task.11 Twenty-one firms providedcompleted responses for the experimental task (fifteen percent of firms contacted).All of the subjects held senior positions within the firms and had undertaken voluntaryadministration engagements. Accordingly, the subjects have considerable experience

in decision-making relevant to the task. Table 11 provides a subject descriptionsummary.

Table 11

Experimental Task Subject Descriptions

Mean Minimum Maximum

Years of insolvency experience 17 8 40

Engagements where reorganisation/ 143 1 1000liquidation has been at issue

Age 44 29 62

Present Position Partner 15

Manager 6

11The experimental task was administered in September 1999.

12No time constraint was imposed for completion of the task, although subjects were advised that it wasestimated the task could be completed in about thirty minutes. Where completion time differed significantlyfrom that suggested it is possible that fatigue or decline in level of interest in the task may have had an

adverse effect on the quality of subjects results. As only a small number of subjects took marginallylonger than the suggested time, the probability of bias due to fatigue or loss of interest was minimal.

The approach to administering the experimental task and instructions provided tosubjects was based on the extensive body of prior behavioural accounting studies(see, for example, Libby 1975 and 1976, Zimmer 1980, and Casey 1980). Each subjectwas provided with financial profiles of twenty real but disguised companies that hadentered voluntary administration between 1993 and 1996. This was a sub-sample ofcompanies included in the statistical model. The smaller but more manageable numberof companies was deemed more appropriate for the experimental task to minimise

problems associated with subjects experiencing information overload or fatigue duringthe task.12Ten of the companies had concluded administration by liquidation; the

other ten had concluded administration with a deed of company arrangement thatprovided for reorganisation of the companys affairs. Of the companies that had


19/27


attempted reorganisation, five were categorised as successful.13Financial profilescontained the same information cues as those used in the statistical model. Data were

provided for two years prior to the year in which the company entered voluntaryadministration.

Subjects were required to identify companies that should attempt reorganisation, andcompanies that should be liquidated. Where subjects identified a company as onethat should attempt reorganisation, they were requested to estimate the probability ofthe company achieving a return on assets that equals or exceeds the industry averagein the first three years after reorganisation. Subjects were also requested to estimatethe probability of the outcome they had indicated, and to rate their confidence in each

prediction. A debriefing questionnaire followed the completion of financial profiles.14

5.1 Predictive Accuracy

In determining subjects prediction accuracy for the performance event, a correctprediction was deemed to have been made when:1. a liquidation or unsuccessful reorganisation was correctly predicted;2. an unsuccessful reorganisation was classified as a liquidation; and3. a successful reorganisation was correctly predicted.This classification determined the ability of subjects to distinguish between companieswith suitable and unsuitable prospects in reorganisation. The accuracy of subjects

prediction in this task is summarised in Table 12.

The binomial theorem indicates that, at the 95 percent confidence level, a subjectpredicts at a rate better than random accuracy if he or she correctly predicts the statusof at least fourteen companies. Table 12 indicates that five subjects performed betterthan random accuracy, correctly predicting the outcome for fourteen or morecompanies. With respect to comparison of the predictive accuracy of the human judgesand the accuracy rates of the statistical model, the average accuracy of human decision-

makers was 10.52 (out of 21 tasks) or 52.61 percent, compared to 87.5 percent for thestatistical model.

Accuracy in identifying companies that were liquidated or were unsuccessful inreorganisation was 59.4 percent. This was nearly double the 32.3 percent accuracyfor identifying companies that successfully reorganised.

13Each subject was advised that (a) the firms had been randomly drawn from a population containingequal proportions of liquidated/reorganised firms, the latter containing 50% deemed successful

reorganisations, and (b) the cost of misclassifying (i) a liquidation/reorganisation and (ii) a successful/unsuccessful reorganisation was the same

14Details of the research instrument are available on request from the authors.


20/27


Table 12

Experts Prediction Success

(Successful/Liquidation and Unsuccessful Reorganisation)

Correct Predictions Overall Number of Subjects (n=21)(20 Cases / 21 Subjects = 442 Predictions)

6 38 49 210 111 412 113 114 315 117 1Correct Predictions = 221 out of 442 (52.61%)Mean = 10.52 (20), Standard Deviation = 3.15,Max = 17, Min = 6

Correct Predictions

Liquidation and Unsuccessful Reorganisation

(15 Cases / 21 Subjects = 315 Predictions)

4 25 16 57 18 110 311 112 413 214 1

Correct Predictions = 187 out of 315 (59.37%)Mean = 8.91 (15), Standard Deviation = 3.29,Max = 14, Min = 4

Correct Predictions

Successful Reorganisation

(5 Cases /21 Subjects = 105 Predictions)

1 72 63 5

Correct Predictions = 34 out of 105 (32.3%)Mean = 1.62 (5), Standard Deviation = 1.02,Max = 3, Min = 0


21/27


Results indicated that experts prediction of successful reorganisation from the cuesprovided was a difficult (if not impossible) task. The results are consistent with thosereported by Houghton and Woodliff (1987) for a task that required classification of

firms based on predicting high or low earnings per share. Their study reported anaverage prediction success of only 54.2 percent.

5.2 Self-Insight into Accuracy of Each Prediction

Subjects were also requested to indicate their degree of confidence in each prediction.The usefulness of financial profile information to subjects is supported if their accuracyis positively associated with confidence in prediction (Zimmer 1980, p.633). Table13 provides the proportion of correct responses for each level of confidence. Overall

the results show little change in accuracy for differing levels of confidence. Thissuggests the information cues were not perceived to be that useful to subjects incarrying out the task. Accurate prediction of successful reorganisations was associatedwith subjects reporting higher levels of confidence, however, the overall accuracywas poor at 20.5 percent.

Table 13

Subjects Correct Response for Confidence Level

(Successful/Unsuccessful Reorganisation)

Prediction Liquidation/Unsuccessful Successful Reorganisation

Confidence Very Confident Not Very Very Confident Not Very

Confident Confident Confident Confident

Fraction 61/96 101/175 25/41 9/44 8/44 1/10Correct

Percentage 63.5% 57.7% 61.0% 20.5% 18.2% 10.0%

CorrectOverall

Very Confident Not Very

Confident Confident

Fraction 26/51 109/219 70/140Correct

Percentage 50.0% 49.8% 51.0%Correct


22/27


5.3 Individual Differences

In an attempt to explain individual differences in prediction accuracy, responses tothe task questionnaire were correlated with prediction accuracy. The correlation matrix

is presented in Table 14. Where it is likely that many variables are correlated withanother variable, potential problems exist with interpreting results of the significanceof univariate correlation as they may not be independent. Accordingly, only correlationsatp


23/27


(6) LIMITATIONS AND CONCLUSION

Data collection for analysis presented in this paper proved difficult and costly. As aresult, the small sample size is a major limitation with respect to the generalisabilityof results for the statistical analysis. The estimation sample may also be subject to

problems of data bias due to the non-random sample selection method. The finalsample of companies may not properly represent the prior probability of reorganisation/

liquidation in the population, which can lead to inconsistent and biased estimates,and result in overstatement of model prediction accuracy (Zmijewski 1984, Piesseand Wood 1992). A further limitation is the reliance on accounting measures that aresubject to bias from managerial manipulations or differences in accounting proceduresor policy choices.

The nature of the experimental task also presents some limitations. The delivery ofthe experimental task may have lacked practical realism, as the financial profiles

provided to subjects may not represent the type of information used by insolvency

experts in assessing a companys prospects. Accordingly, results from the experimentwould not be completely generalisable to other situations.

Overall, the results of this exploratory study have several implications. The maincontribution of this paper is the insight it provides regarding the relevance of financialinformation to VA decision-making.

With respect to the reorganisation decision for companies that enter VA, our analysisindicates that financial characteristics differ between companies that reorganise andcompanies that proceed to liquidation. The ability to distinguish betweenreorganisations and liquidations is useful for stakeholders whose costs systematically

Table 15

Summary: Subjects Usefulness Rating of Information Cues

Information Mean Std. Deviation Range

Cue (10 point scale) Low HighCACL 7.881 2.156 1 10

OPTA 7.452 1.774 4 10

TA 7.214 2.171 2.5 10

TLTA 6.929 2.087 1 10

INDUSTRY 6.214 1.586 2 8

CLTA 5.500 2.388 1 10

TLOE 5.119 2.155 2 10


24/27


differ between the two outcomes. Significant variables in the reorganisation eventmodel may reflect the existence of bargaining or coalition behaviour betweenstakeholders prior to the reorganisation event. Future research might focus oninvestigation of the reorganisation event with a view to gaining a greater understanding

of how financial position affects decision-making.

The analysis presented also indicates that financial characteristics differ betweencompanies that have successful and unsuccessful reorganisation outcomes. Statisticalanalysis conducted indicates that success in reorganisation is associated with higherlevels of past profitability and short-term liquidity.

In comparing the predictive performance of the statistical analysis and human decision-makers we found that, although the statistical model had high classification accuracy,

the decision accuracy of insolvency experts in an experimental task (using the samefinancial variables) was poor. The results indicated that experts prediction of successfulreorganisation from the financial cues provided was a difficult (if not impossible)task. A limitation of this comparison arises because the statistical models are developedusing information about actual outcomes, while the expert decision-makers were only

provided with information about prior probabilities of the outcomes. However, theresult suggests that the experts use of information cues is less than optimal, and thatutilization of statistical models may improve the performance of decision-making.While the problem of deciding whether a company should attempt reorganisation isnot likely to be reduced solely to consideration of financial variables, our findingssuggest that it might be advantageous to employ an appropriate statistical model toassist those required to make decisions or give advice on company reorganisation.Furthermore, informed decision-making in VA may contribute to minimising theincidence of prolonging the existence of non-viable firms, thereby reducing costsassociated with the VA insolvency regime. Future research might address thedevelopment and refinement of suitable models.

REFERENCES

Abdel-khalik, A.R. and K.M. El-Sheshai (1980). Information Choice and Utilisation in an Experiment of

Default Prediction,Journal of Accounting Research, 18, 325-342.

Altman, E.I. (1968). Financial Ratios, Discriminant Anlaysis and the Prediction of Corporate Bankruptcy.

Journal of Finance, 23, 589-609.

Altman, E.I. (1971). Corporate Bankruptcy in America, Heath Inc, Lexington.

Altman, E.I. (1984). A Further Empirical Investigation of the Bankruptcy Cost Question.Journal of Finance,

39, 1067-1089.

Altman, E.I. and T.P. McGough (1974). Evaluation of a Company as a Going Concern. Journal of

Accountancy, December, 50-57.


25/27


Ang, J.S. and J.H. Chua (1980). Coalitions, The Me-First Rule, and the Liquidation Decision.Bell Journal

of Economics, 1, 355-359.

Australian Securities and Investment Commission (ASIC), (1998). A Study of Voluntary Administration

in New South Wales, ASIC Research paper 98/01.

Barniv, R., Agarwal, A. and R. Leach (2002). Predicting Bankruptcy Resolution. Journal of Business

Finance and Accounting, 29, 497-520.

Brunswik, E. (1952). The Conceptual Framework of Psychology, in Carnap, R. and Morris, C. (eds),

International Encyclopaedia of Unified Science, University of Chicago Press, Chicago.

Bulow, J.I. and J.B. Shoven (1978). The Bankruptcy Decision.Bell Journal of Economics, 9, 437-456.

Campbell, S.V. (1993). The Significance of Direct Bankruptcy Costs in Determining the Outcome of

Bankruptcy Reorganization. Unpublished Doctoral Dissertation, University of Oregon.

Campbell, S.V. (1996). Predicting Bankruptcy Reorganization for Closely Held Firms.Accounting Horizons,

10, 12-25.

Casey, C.J. (1980). Additional Evidence on the Usefulness of Accounting Ratios for Subjects Predictions

of Corporate Failure.Journal of Accounting Research, 18, 603-613.

Casey, C.J. (1983). Prior Probability Disclosure and Loan Officers Judgements: Some Evidence of the

Impact.Journal of Accounting Research, 21, 300-307.

Casey, C.J. McGee, V.E. and C.P. Stickney (1986). Discriminating Between Reorganized and Liquidated

Firms in Bankruptcy. The Accounting Review, 61, 249-262.

Castagna, A.D. and Z.P. Matolcsy (1981). The Prediction of Corporate Failure: Testing The Australian

Experience.Australian Journal of Management, 6, 23-50.

Casterella, J.R., Lewis, B.L. and P.L. Walker (2000). Modelling Audit Opinions Issued to Bankrupt

Companies: A Two-Stage Empirical Analysis.Decision Sciences, 31, 507-530.

Comerford, R.A. (1976). Bankruptcy as a Business Strategy: A Multivariate Analysis of the Financial

Characteristics of Firms Which Have Succeeded in Chapter XI Compared to Those Which Have Failed,

Unpublished Doctoral Dissertation, University of Massachusetts.

Crutchfield, P. 1994,Annotated Corporate Voluntary Administration Law, The Law Book Company, Sydney.

Cybinski, P.J. (1995). A Discrete-Valued Risk Function For Modelling Financial Distress in Private

Australian Companies.Accounting and Finance, 35, 17-32.

Deakin, E.B. (1977). Business Failure Prediction: An Empirical Analysis, in Altman, E.I. and Sametz,

A.W. (eds), Financial Crisis; Institutions and Markets in a Fragile Environment, John Wiley and Sons,

72-88.

Fisher, T.C. and J. Martel (1995). The Creditors Financial Reorganisation Decision: New Evidence From

Canadian Data.Journal of Law, Economics and Organization, 11, 112-126.

Frost-Drury, A., Greinke, A., and G. Shailer (1998). Financial Differences Between Voluntary

Administrations and Liquidations.Insolvency Law Journal, 6, 153-158.


26/27


Frost-Drury, A., Greinke, A., and G. Shailer (2000). Distinguishing Distressed Companies Choosing

Voluntary Administration.Accounting, Accountability and Performance, 6, 19-31.

Gertner, R. and D. Scharfstein (1991). A Theory of Workouts and the Effects of Reorganization Law.

Journal of Finance, 46, 1189-1222.

Gilbert, L.R., Menon, K. and K.B. Schwartz (1990). Predicting Bankruptcy for Firms in Financial Distress.

Journal of Business Finance and Accounting, 17, 161-171.

Hodson, G., and D. McEvoy (1995). Voluntary Administrations: Are They Really Working? Coopers &

Lybrand - Business Imperatives Series No. 1.

Hotchkiss, E.S. (1995). Postbankruptcy Performance and Management Turnover.Journal of Finance, 50,

3-21.

Houghton, K.A. (1984). Accounting Data and the Prediction of Business Failure: The Setting of Priors and

the Age of Data.Journal of Accounting Research, 22, 361-368.

Houghton, K.A. and R. Sengupta (1984). The Effect of Prior Probability Disclosure and Information Set

Construction on Bankers Ability to Predict Failure,Journal of Accounting Research, 22, 768-775.

Houghton, K.A. and D.R. Woodliff (1987). Financial ratios: The Prediction of Corporate Success and

Failure.Journal of Business Finance and Accounting, 14, 537-554.

Izan, H. (1984). Corporate Distress in Australia.Journal of Banking and Finance, 8, 303-320.

John, K. (1993). Managing Financial Distress and Valuing Distressed Securities: A Survey and Research

Agenda. Financial Management, 22, 60-86.

Kennedy, D.B. and W.H. Shaw (1991). Evaluating Financial Distress Resolution Using Prior Audit Opinions.

Contemporary Accounting Research, 8, 97-114.

Kida, T. (1980). An Investigation into Auditors Continuity and Related Qualification Judgements.Journal

of Accounting Research, 18, 506-523.

Koh, H.C. (1991). Model Predictions and Auditor Assessments of Going Concern Status.Accounting and

Business Research, 21, 331-338.

Lachenbruch, P. (1975).Discriminant Analysis, Hafner, New York.

Lenard, M.J., Alam P. and G.R. Madey (1995). The Application of Neural Networks and a Qualtitative

Response Model to the Auditors Going Concern Uncertainty Decision.Decision Sciences, 26, 209-227.

Libby, R. (1975). Accounting Ratios and the Prediction of Failure: Some Behavioural Evidence.Journal

of Accounting Research, 13, 150-160.

Levy, A. (1992). A Pareto Optimal Collaboration Period: The Role of Financial, Industrial and

Macroeconomic Conditions in Liquidation Decision and Timing.Journal of Economics and Finance, 16,

1-12.

Levy, A. and R. Bar-Niv (1987). Macroeconomic Aspects of Firm Bankruptcy Analysis. Journal of

Macroeconomics, 9, 407-415

Libby, R. (1976). Man Versus Model of Man: Some Conflicting Evidence. Organizational Behaviour and

Human Performance, 16, 1-12.


27/27


Lincoln, M. (1984). An Empirical Study of the Usefulness of Accounting Ratios to Describe Levels of

Insolvency Risk. Journal of Banking and Finance, 8, 321-340.

LoPucki, L.M. (1983). The Debtor in Full Control - Systems Failure Under Chapter 11 of the Bankruptcy

Code? (first instalment).American Bankruptcy Law Journal, 57, 99-116.

LoPucki, L.M. (1983). The Debtor in Full Control - Systems Failure Under Chapter 11 of the Bankruptcy

Code? (second instalment).American Bankruptcy Law Journal, 57, 247-273.

Martel, J. (1991). Bankruptcy Law and the Canadian Experience: An Economic Appraisal. Canadian

Public Policy, 17, 52-63.

McNamara, R.P., Cocks, N.J. and D.F. Hamilton (1988). Predicting Private Company Failure.Accounting

and Finance, 28, 53-64.

Piesse, J. and D. Wood (1992). Issues in Assessing MDA Models of Corporate Failure: A Research Note.

British Accounting Review, 24, 33-42.

Pinches, G.E., Mingo, K.A. and J.K. Caruthers (1977). The Stability of Financial Patterns in Industrial

Organizations. The Journal of Finance, 28, 389-396.

Rose, P.S., Andrews, W.T. and G.A. Giroux (1982). Predicting Business Failure: A Macroeconomic

Perspective.Journal of Accounting, Auditing and Finance, 6, 20-31.

Shailer, G.E.P. (1990). The Multivariate Performance of Alternative Accounting Variables for Predicting

Unlisted Company Failures.British Accounting Review, 22, 151-162.

Simnett, R. and K. Trotman (1989). Auditor Versus Model: Information Choice and Information Processing.

The Accounting Review, 64, 514-528.

Ward, T.J. and B.P. Foster (1997). A Note on Selecting A Response Measure For Financial Distress,Journal

of Business Finance and Accounting, 24, 869-879.

Warner, J. (1977). Bankruptcy Costs: Some Evidence.Journal of Finance, 32, 337-347.

White, M. (1980). Public Policy Toward Bankruptcy: Me-first and Other Priority Rules. The Bell Journal

of Economics, 11, 550-564.

White, M. (1983). Bankruptcy Costs and the New Bankruptcy Code.Journal of Finance, 38, 477-488.

White, M. (1984). Bankruptcy, Liquidation and Reorganization inHandbook of Modern Finance, Logue,D. (Ed), Warren, Gorham & Lamont, New York.

White, M. (1989). The Corporate Bankruptcy Decision.Journal of Economic Perspectives, 3, 129-151.

Zavgren, C.V. (1985). Assessing the Vulnerability to Failure of American Industrial Firms: A Logistic

Analysis.Journal of Business Finance and Accounting, Spring, 19-45.

Zimmer, I. (1980). A Lens Study of the Prediction of Corporate Failure by Bank Loan Officers.Journal of

Accounting Research, 18, 629-636.

Zimmer, I. (1981). Modelling Lenders Assessments of the Ability of Corporate Borrowers to Repa.Abacus,17, 145-160.

optiuneva+reorganizare+falim_australia

Documents