USING RATIOS TO PREDICT FINANCIAL FAILURE
Financial ratios, based on current or past performance, are often used to help predict the future. Normally, both the choice of ratios and the interpretation of results are depend- ent on the judgement and opinion of the analyst. However, there have been attempts to develop a more rigorous and systematic approach to the use of ratios for prediction pur- poses. These attempts have often focused on the ability of ratios to predict the financial failure of a business.
By financial failure, we mean a business either being forced out of business or being severely adversely affected by its inability to meet its financial obligations. It is often referred to as ‘going bust’ or ‘going bankrupt’. This, of course, is an area of concern for all those connected with the business.
Using single ratios
Various methods of using ratios to predict future financial failure have been developed. Early research looked at whether a single ratio was a good or bad predictor of financial failure. It involved tracking a particular ratio (for example, the current ratio) for a business over several years leading up to the date of the failure. This was to see whether the ratio exhibited a trend that could be taken as a warning sign.
Beaver (see reference 1 at the end of the chapter) carried out the first research in this area.
He identified 79 businesses that had failed. He then calculated the average (mean) of various ratios for these 79 businesses, going back over the financial statements of each business for each of the ten years leading up to each business’s failure. Beaver then compared these aver- age ratios with similarly derived ratios for a sample of 79 businesses that did not fail over this period. (The research used a matched-pair design, where each failed business was matched with a non-failed business of similar size and industry type.) He found that some ratios exhib- ited a marked difference between the failed and non-failed businesses for up to five years prior to failure. These were:
■ cash flow/total debt
■ net income (profit)/total assets
■ total debt/total assets
■ working capital/total assets
■ current ratio
■ no credit interval (that is, cash generated from operations to maturing obligations).
To illustrate Beaver’s findings, the average current ratio of failed businesses for five years prior to failure, along with the average current ratio of non-failed businesses for the same period, is shown in Figure 3.15.
Sainsbury. Sainsbury’s current ratio shows a fairly consistent downward path until 2010.
Morrison has tended to maintain the lowest current ratio over time. With well-managed businesses like these, it is highly likely that any changes are the result of deliberate policy.
Source: Annual reports of the three businesses 2008 to 2018.
www.freebookslides.com
128 CHAPTER 3 ANALYSING AND INTERPRETING FINANCIAL STATEMENTS
Let us assume that research indicates that a particular ratio is shown to be a good predic- tor of failure. Can you think of a practical problem that may arise when using this ratio to predict financial failure for a particular business?
Where a particular ratio for a business differs from the mean ratios of non-failed businesses, the analyst must rely on judgement to interpret whether it is significant. There is no clear decision rule that can be applied. Different analysts may therefore come to different conclu- sions about the likelihood of failure.
Activity 3.27
Figure 3.15 Average (mean) current ratio of failed and non-failed businesses
The vertical scale of the graph is the average value of the current ratio for each group of businesses (failed and non-failed). The horizontal axis is the number of years before failure. Thus, Year 1 is the most recent year and Year 5 the least recent year. We can see that a clear difference between the average for the failed and non-failed businesses can be detected five years prior to the failure of the former group.
3.5
3.0
Current 2.5 ratio
2.0
0 5 4 3
Number of years before failure Failed businesses
Non-failed businesses
2 1
Research by Zmijewski (see reference 2 at the end of the chapter), using a sample of 72 failed and 3,573 non-failed businesses over a six-year period, found that businesses that ultimately went on to fail were characterised by lower rates of return, higher levels of gearing, lower levels of coverage for their fixed interest payments and more variable returns on shares.
While we may not find these results very surprising, it is interesting to note that Zmijewski, like a number of other researchers in this area, did not find liquidity ratios particularly useful in predicting financial failure. Intuition might have led us (wrongly it seems) to believe that the liquidity ratios would have been particularly helpful in this context. As we saw earlier, however, Beaver did find the current ratio to be a useful predictor.
The approach adopted by Beaver and Zmijewski is referred to as univariate analysis because it looks at one ratio at a time. It can produce interesting results, but there are practi- cal problems associated with its use.
www.freebookslides.com
USING RATIOS TO PREDICT FINANCIAL FAILURE 129
A further problem arises where more than one ratio is used to predict failure. Let us say, for example, that past research has identified two ratios as being good predictors of finan- cial failure. When applied to a particular business, however, it may be that one ratio predicts financial failure, whereas the other does not. Given these conflicting signals, how should the analyst interpret the results?
Using combinations of ratios
The weaknesses of univariate analysis have led to the development of models that combine ratios so as to produce a single index that can be interpreted more clearly. One approach to model development, much favoured by researchers, uses multiple discriminate analysis (MDA). This is, in essence, a statistical technique that is similar to regression analysis and which can be used to draw a boundary between those businesses that fail and those businesses that do not. This boundary is referred to as the discriminate function. In this context, MDA attempts to identify those factors likely to influence financial failure. MDA differs from regres- sion analysis in that it assumes that the observations come from two different populations (for example, failed and non-failed businesses) rather than from a single population.
To illustrate this approach, let us assume that we wish to test whether two ratios (say, the cur- rent ratio and the return on capital employed) can help to predict failure. To do this, we can calcu- late these ratios, first for a sample of failed businesses and then for a matched sample of non-failed ones. From these two sets of data we can produce a scatter diagram that plots each business according to these two ratios to produce a single co-ordinate. Figure 3.16 illustrates this approach.
The distribution of failed and non-failed businesses is based on two ratios. The line represents a boundary between the samples of failed and non-failed businesses. Although there is some crossing of the boundary, the boundary represents the line that minimises the problem of misclassifying particular businesses.
Figure 3.16 Scatter diagram showing the distribution of failed and non-failed businesses
Current ratio
ROCE ratio Failed businesses Non-failed businesses
Using the observations plotted on the diagram, we try to identify the boundary between the failed and the non-failed businesses. This is the diagonal line in Figure 3.16. We can see that those businesses that fall below and to the left of the line are predominantly failed and those that fall to the right are predominantly non-failed ones. Note that there is some overlap
www.freebookslides.com
130 CHAPTER 3 ANALYSING AND INTERPRETING FINANCIAL STATEMENTS
between the two populations. In practice, the boundary produced is unlikely to eliminate all errors. Some businesses that fail may fall on the non-failed side of the boundary. The opposite also happens. However, the analysis will minimise the misclassification errors.
The boundary shown in Figure 3.16 can be expressed in the form:
Z = a + (b : Current ratio) + (c : ROCE)
where a, b and c are all constants and b and c are weights to be attached to each ratio.
A weighted average or total score (Z) is then derived. By ‘constants’ we mean that the same values are used for assessing each individual business. The values ascribed to these constants are those that have been found in practice to provide a Z-score that most effectively is able to differentiate between the failed and the non-failed businesses. Using this model to assess a particular business’s health, we would deduce the current and ROCE ratios for that business and use them in the equation above. If the resulting Z-score were to come out below a certain value, we should view that business as being at risk.
Note that this example, using the current and ROCE ratios, is purely hypothetical and intended only to illustrate the approach.
Z-score models
Altman was the first to combine financial ratios in a way that successfully predicted financial distress among businesses. The Z-score model that he devised was developed in 1968 and is based on five financial ratios. The Z-score model is as follows:
Z = 1.2a + 1.4b + 3.3c + 0.6d + 1.0e where:
a = Working capital (current assets – current liabilities)/Total assets b = Retained earnings/Total assets
c = Operating profit/Total assets
d = Market value of equity shares/Total liabilities at book (statement of financial position) value e = Sales revenue/Total assets
The weightings (or coefficients) in the above model are constants that reflect the importance to the Z-score of each of the ratios (a to e). It is interesting to note that Operating profit/Total assets (a profitability ratio) is given far more weight in the model than Working capital/Total assets (a liquidity ratio). The five ratios employed were identified by Altman through trial and error, as there is no theory of financial distress to offer guidance as to which ratios should be chosen.
According to Altman, those businesses with a Z-score of less than 1.81 occupy a ‘distress zone’. This means that they are unstable and have a high risk of failure within two years. (The lower the score, the greater the risk of failure.) Those with a Z-score greater than 2.99 are considered to be financially stable and occupy a ‘safe zone’. Those businesses with a Z-score between 1.81 and 2.99 occupy a ‘grey zone’, where there is some risk of financial distress within two years. The model has demonstrated a high level of predictive ability. In a series of tests over sample periods up to 1999, it had around 80 per cent and 90 per cent accuracy in predicting financial distress one year prior to the event. However, the accuracy of the model diminishes significantly as the lead time increases (see reference 3 at the end of the chapter).
Altman based his original model on US manufacturing businesses but later adapted it to non-manufacturing businesses. (The adapted model excludes the Sales revenue/Total assets ratio as non-manufacturing businesses tend to have a smaller asset base.) Further adaptations were carried out to accommodate private businesses and businesses in emerging markets.
www.freebookslides.com
USING RATIOS TO PREDICT FINANCIAL FAILURE 131
In recent years, other models, using a similar approach, have been developed throughout the world. In the UK, Taffler has developed separate Z-score models for different types of business. (See reference 4 at the end of the chapter for a discussion of the work of Taffler and others.)
Real World 3.13 shows the distribution of Z-scores among large US businesses.
Knowing the score
A study of the 500 largest listed US businesses was carried out, which revealed the distribu- tion of Z-scores shown in Figure 3.17.
We can see that the distribution (after taken out six extreme observations) is skewed. Most have a score of 3.0 or more, which indicates that they are not in danger of financial distress.
This distribution is as we might expect from such large, established, companies.
Source: Li, J. (2016) ‘Tech companies reach high Altman z-scores’, gurufocus.com, 14 July.
Real World 3.13
Figure 3.17 Distribution of Z-score for S&P 500 companies (Outliers removed) The distribution is skewed to the right as we might expect.
100 120
80
60
40
Frequency
20
0 –3 –2 –1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Z-Score
There has been a great deal of investigation, much of it by academics, into the effectiveness of Altman’s Z-score model, or some variation of it. These investigations, which span different economies and industries, overwhelmingly support the model. Further support can be seen through its widespread use among businesses and independent analysts for decision-making purposes.
The prediction of financial failure is not the only area where research into the predictive ability of ratios has taken place. Researchers have also developed ratio-based models to assess the vulnerability of a business to takeover by another. This is another area that is of vital importance to all those connected with the business.
www.freebookslides.com
132 CHAPTER 3 ANALYSING AND INTERPRETING FINANCIAL STATEMENTS