ABSTRACT
The continuing search for a procedure which will help analysts to identify companies that are likely to go bankrupt is understandable. The simplest procedure is to develop models which include a number of financial ratios, the parameter weights being adapted to improve discriminatory power. The share price studies largely confirmed what other studies have shown, namely that on average the market appears to distinguish between failing and sound firms some 2-3 years before their final year end date. Further light on the difficulties of trying to discriminate between failing and non-failing businesses was shed by the 25 case studies undertaken. A major problem with all these modelling techniques is that they are typically applied to heterogeneous financial data relating to companies drawn from several disparate industry categories. The data are almost always pooled across time when economic circumstances are changing. The behaviour of share prices suggests that such models may anyway largely be imitating the procedures already employed by analysts.
