Shortcomings of the Bankruptcy Prediction Models

In the previous article, we have studied how bankruptcy prediction models have come a long way. It is true that they help investors make an educated guess when they put their money in a company. However, it is also important to understand that these methods are nowhere close to infallible. In fact, these models have a lot of shortcomings that investors must be aware of while making decisions based on interpretations of such models.

In this article, we have listed down some of the common shortcomings commonly reported by analysts while using bankruptcy prediction models.

Uses Data From Financial Statements

Bankruptcy prediction models use data from financial statements. This data may not be accurate a lot of the time. This is because it is a known fact that companies maintain multiple sets of books. One set of books is maintained with an objective to lower the tax liabilities. On the other hand, a different set of books is maintained to impress investors. This is the set of books which is commonly available to the public. The accounting assumptions used in these statements are quite liberal and are meant to display the company in a good light. Hence using revenues, retained earnings, and other numbers from these statements may not give the correct picture. If the same model is run using data from a different set of books, the result may be quite different!

Do Not Pay Much Attention to Cash Flow

Another important point to note is that very few bankruptcy prediction models pay much attention to cash flow. This is absurd given the fact that bankruptcy is actually a situation when the company does not have adequate cash flow. However, none of the ratios used in Altman’s Z score model or Merton’s model specifically measure cash flow.

The reality is that every number on the accounting statement can be fudged except cash. Hence, when bankruptcy models do not use cash in their analysis, they invite more shortcomings into their analysis.

Models May be Industry Specific

Univariate models for bankruptcy prediction tend to look at only one variable. This variable is often defined as keeping particular industries in mind. Therefore, these models are industry-specific. However, in practice, analysts use these models across industries. It is, therefore, important to realize that in such cases, the model may not give the right result.

Coefficients Need to be Updated Because of Time

Many models like the Altman Z score rely on coefficients. These coefficients provide the weight to the particular ratio. However, it is important to note that the data on which the Altman model is based was collected in the 1960s and 1970s. It is therefore plausible to assume that the data has now become outdated and that the companies must update the coefficients. A lot has changed since the 1970s, and these changes do affect the solvency of any firm. There is a dire need to collect new data, conduct an elaborate empirical analysis, and change the coefficients in the Altman model. Many companies are aware of this shortcoming and hence use a modified version of the Altman model for their analysis.

Coefficients Need To Be Updated Since Industries Have Changed

It also needs to be understood that the data on which the Altman Z model has been created was collected from companies in the manufacturing domain. Business models have undergone a transformation since that time. These models were developed when tech companies like Google, Microsoft, and Facebook did not even exist. It would, therefore, be fair to say that these models are not applicable to new generation tech companies, which include many promising start-ups as well as behemoths like the ones mentioned above.

Many of these companies follow unconventional financial models. For instance, companies like Facebook and Twitter did not have retained earnings for a very long time. Also, they had a negative cash flow for many years. The old models, such as Altman’s Z score, are not capable of analyzing these companies. A newer model needs to be developed on the lines of Altman’s Z model.

Not Applicable to Financial Firms

Bankruptcy prediction models are not applicable to financial firms either. This is unbelievable since the Great Recession of 2008 was caused by the bankruptcy of financial institutions. These institutions form the backbone of the economy, and hence any problems in them, lead to a systemic crisis.

However, financial firms tend to have a complicated and opaque balance sheet. There is significant use of off-balance-sheet accounting in these firms. It is for this reason that they are not suitable for analysis in models which are based on the balance sheet.

Cannot Predict Outliers

Another important lesson from the 2008 crisis is that outliers can cause a lot of damage. Outliers are companies that have a low probability of default, but if the default actually happens, then the impact is very high. Common bankruptcy prediction models do not have any mechanism to conduct a special analysis of the outliers. Therefore, newer models are required which pay special attention to institutions whose bankruptcy can cause high impact and can shock the entire ecosystem.

To sum it up, the bankruptcy models are not anywhere close to perfect. However, not using them isn’t an option either. Hence, it is important for the investor to be aware of when these models aid in decision making and when they don’t!