Bankruptcy prediction models Hughes, Stewart. Credit Control. Hutton: 1993. Vol. 14, Iss. 11; pg.

16, 7 pgs Abstract (Summary) Bankruptcy prediction models are reviewed. The relative merits of univariate financial ratio analysis, multi-discriminant analysis, and managerial performance models are examined. It is found that each of these models needs to be supplemented by a sobering review of economic fundamentals. Financial analysts and experienced credit managers are often able to predict the likelihood of a firm experiencing financial difficulties, possibly leading to bankruptcy, by monitoring closely its performance over a long period. Techniques available for predicting corporate bankruptcy are discussed. In general, if one is in the business of analyzing company performance, it is clear that both financial and non-financial indicators should be taken into account. Moreover, it is inappropriate simply to use historic accounting data and project it into the future. It is necessary to incorporate into the analysis predictions about the level of economic activity and, in particular, the impact of prolonged high interest rates. Full Text (1789 words) Copyright House of Words, Ltd. 1993 With the British economy emerging from a deep recession, it is an appropriate moment to review the use of bankruptcy prediction models. Here we examine the relative merits of univariate financial ratio analysis, multi-discriminant analysis and managerial performance models. We find that each of these models needs to be supplemented by a sobering review of economic fundamentals. Financial analysts and experienced Credit Managers are often able to predict the likelihood of a firm experiencing financial difficulties, possibly leading to bankruptcy, by monitoring closely its performance over a long period. Techniques available for predicting corporate bankruptcy fall into the three major types discussed below. 1. UNIVARIATE FINANCIAL RATIO ANALYSIS The analysis of a company's financial ratios both on a cross-sectional basis (comparisons with firms in similar industries) and on a time-series basis (comparisons within the same firm over time) have been the traditional method used to predict deterioration in a company's financial health. The major ratios used in this kind of analysis have included profitability ratios, liquidity ratios, gearing ratios, activity ratios and investment ratios. Deterioration in these either over time within the same company or in respect to intercompany comparisons frequently set alarm bells ringing in the boardroom and among the financial press. The problems of using such ratios for performance prediction are well documented, particularly those involving inter-company comparisons in industries where fins are highly diversified on both a product and geographical basis. However, besides these problems, from the point of view of bankruptcy prediction, there arises the basic problem that financial ratio analysis is essentially a univariate technique. In other words, the

problem with financial ratio analysis is that each ratio is examined separately, in isolation from the other ratios. Consequently, the analysis has the problem of not only deciding whether a particular ratio is good or bad (e.g., a low gearing ratio could indicate either that there is a strong equity base or that the firm is reluctant to take on risk) but also that different ratios may be moving in opposite directions, thereby indicating different predictions. In effect, the major problem of using univariate financial ratio analysis for bankruptcy prediction purposes is that the combined effect of several ratios are based solely on the subjective judgement of the financial analyst. 2. MULTIPLE DISCRIMINANT ANALYSIS (MDA) This was developed largely in response to the shortcomings of univariate financial ratio analysis. Multi-discriminant analysis (MDA) starts from the premise that a company's accounts are multivariate documents in that they measure several aspects of a company's performance simultaneously. Therefore, to evaluate a company simply on the basis of one aspect of its performance such as its profitability would be dangerously to ignore other aspects of its performance such as its liquidity and its financial risk. The first attempt to use MDA for bankruptcy prediction purposes was the model developed by Edward Altman in the US in 1968. Since then the technique has been developed by Taffler, Tisshaw and others in the United Kingdom. Essentially, all variations of the technique involve the following three steps: (a) establish two mutually exclusive groups, namely those firms which have gone bankrupt and those which are still continuing to trade successfully; (b) collect financial ratios for both of these two groups; (c) identify the financial ratios which best discriminate between the two groups. Altman, in his original work, found that there were five key ratios which best discriminated between the bankrupt and non-bankrupt firms. He used these ratios to calculate a Z score as follows: Z=.012X sub 1 +.014X sub 2 +.033X sub 3 +.006X sub 4 + .999X sub 5 where X sub 1 =working capital/total assets (%) X sub 2 =retained earnings/total assets (%) X sub 3 =earnings before interest and tax/total assets (%) X sub 4 =market value of equity/book value of debt(%) X sub 5 =sales/total assets

Altman applied his model to 66 firms, half of which had gone bankrupt. He found that the average of the five ratios for his equal groups of bankrupt and non-bankrupt firms were as follows: (Averages omitted) Substituting these values into the equation, the Z score for bankrupt and non bankrupt firms becomes: (Z score omitted) Altman found that the key Z score was 2.675. Firms with Z scores above this were declared as non-bankrupt while those with Z scores below this were classified as bankrupt. The most important discriminator between bankrupt and non-bankrupt firms he found to be the profitability ratio X sub 3 , followed by X sub 5 , X sub 4 , X sub 2 and X sub 1 . He further claimed that using a Z score of 2.675 as the classification boundary (i.e., as between bankrupt and non-bankrupt firms) that his model was 95 percent accurate in predicting one year prior to bankruptcy, and 72 percent accurate when applied to company data two to five years prior to bankruptcy. MDA analysis has clear advantages over simple financial ratio analysis in that it is based on objective statistical data rather than upon the subjective interpretations of the financial analyst. However, any model can only be as good as the data on which it is based. Consequently, even assuming the predictive accuracy of MDA models are as good as their exponents argue, there are still the problems of changes in accounting practices, creative accounting and the fact that companies in financial difficulties frequently tend to delay the publication of their accounts. Moreover, MDA models are based on historical data and may fail to identify major changes in the company's competitive environment or management structure. Thus, predictions based on MDA may be useful starting points but they are unlikely to tell the whole story. 3. MANAGERIAL PERFORMANCE MODELS Managerial models are much more subjective than MDA models and are based upon the analyst's judgment in relation to the overall managerial, financial and trading position of the firm. The best known of these models is the Argenti 'A' score model in which he attempts to quantify performance by attaching scores to various characteristics of performance. Scores are awarded under three major headings in the Argenti framework, namely defects, mistakes and symptoms. A maximum of 100 marks may be awarded overall, comprising 43 for the defects section, 45 for mistakes and 12 for symptoms. The higher the score awarded the more likely it is the company is badly run and is heading for failure. In the Argenti model, defects are deemed to be of three major types. The first is the managerial structure where Argenti argues that failure is most likely to be associated with autocratic chief executives, particularly where the chief executive is also the chairman. If

is accompanied by an unbalanced, passive board with a weak finance director and a lack of professional managers below board level, the probability of failure is likely to increase. The second defect is to be found in weak accounting systems, particularly where there is no proper budgetary control system, an inadequate cash flow planning system and poor or nonexistent product costing. The final defect is to be found in management's lack of response to change, particularly in relation to changing products, processes, markets and work practices. Of the 43 marks awarded for the defects section, 19 are awarded for management structure, 9 for accounting controls, and 15 for responsiveness to change. Because of these defects in the managerial and accounting set-up, Argenti argues that these can lead to three major mistakes being made. The first of these is overtrading, whereby the company's turnover rises faster than its cash availability, thereby leading to cash flow problems. The second is where the company's financial structure becomes characterised by high gearing, so that the interest on the company's loans become a major burden on its profits. The third major mistake is the big project, where the company takes on a project of such a scale in relation to the company's size, that if the project goes wrong it can cause the entire company to collapse. Each of these mistakes are regarded as of equal magnitude in the Argenti model and are each awarded 15 marks in his scoring system. The third section of Argenti's model concerns the symptoms which begin to appear as the company lurches towards failure. (Argenti's model omitted). These include both financial and non-financial symptoms. Financial symptoms take the form of deteriorating financial ratios or Z scores and the use of creative accounting which together are awarded eight marks. Finally, nonfinancial indicators such as declining morale, market share, adverse rumours and resignations together comprise the final four marks. The scoring system as devised by Argenti is extremely rigid. The allocation of marks is an all or nothing procedure, with either the full mark being awarded or a zero mark, the model not allowing any intermediate scores. The overall danger mark above which companies may be in danger of failing is 25, although an individual score of 10 or more for the defects and 15 for the mistakes would also put the company at serious risk. There is plenty of scope for developing models along the Argenti line of thinking. The problem with these models is that they tend to be based upon subjective judgements, not only in terms of the variables to be included in the model but also in terms of the scoring system to adopt. It is an arbitrary judgment as to what is the appropriate level of gearing or whether existing management is autocratic. An interesting feature of the Argenti model is that it relegates deteriorating financial ratios to a relatively minor role. However, casual empiricism tends to lend support to the major role played by the three big mistakes of overtaking, overgearing and the big project, in causing company collapse. What, then, can we conclude about bankruptcy prediction models? If one is in the business of analysing company performance, it is clear that both financial and nonfinancial indicators should be taken into account. Moreover, it is inappropriate simply to

use historic accounting data and project it into the future. It is necessary to incorporate into the analysis, predictions about the level of economic activity and in particular, the impact of prolonged high interest rates. The inability to predict some of the major corporate collapses of the past two years suggests that many analyst are still not taking account of the impact of economic fundamentals upon future performance. REFERENCES A.I. Altman: 'Discriminant analysis and the prediction of corporate bankruptcy", Journal of Finance, September 68, Vol 23. J. Argenti: 'Discerning the signs of company failure", The Director, October 83, Vol 37. This article first appeared in the June 1993 issue of Student Digest and is reproduced with the kind permission of the Institute of Company Accountants. Stewart Hughes, MA, FSCA, Senior Lecturer, Coventry University