Models of bankruptcy prediction – the discriminant analysis
With the balance sheet dataset, we are able to run a discriminant analysis that helps to estimate the bankruptcy threat. Discriminant analysis (DA) relies on the predictive power of the discriminant function (Z). The discriminant models apply the set of financial ratios that reflect various aspects of a company's performance and financial position. Grounded on the results of the procedure of discriminant analysis (comparison of groups of constrained and unconstrained companies), the model provides the relevant discriminant ratios and their wages in the model.
Discriminant analysis is very useful in the evaluation of the overall financial strength and position of companies that aim to obtain additional external funding (e.g. bank loan).
One of the most widely recognized models is Altman's Z-score which was developed in 1968. for companies operating in the US. However, it is highly recommended to rely on the DA models that were developed for companies that perform in similar economic conditions which are determined by the country of operating activity (in this case – Poland). It is also highly recommended to use simultaneously several DA models for bankruptcy prediction (3-5) for greater accuracy of the prediction of real bankruptcy threat.
The group of models of bankruptcy prediction includes:
- A. Holda's model (2001),
- The "Poznanski" model (2004),
- E. Maczynska's model,
- B. Prusak's model,
- D. Hadasik's model (1998).
Apart from the above models, the risk of bankruptcy can be also assesed using Wilcox ratio, which defines the company's ability to repay the current debts, as well as the ability to cover the potential future liabilities.