ABSTRACT

An alternative to using multiple discriminant analysis is the use of conditional probability models to estimate the probability of occurrence of a choice or outcome, conditional on the attribute vector of the choice or outcome set that is available. The second half of this chapter will show that although conditional probability models have a number of advantages over multiple discriminant analysis they are generally unable to reduce the multi-dimensionality of the problem of corporate failure. This implies that either a theory or some form of dimensionality reduction technique will have to be employed to ensure that the variable choice set is of the right form before using the conditional probability models. The earlier chapters argued that the development of theory leaves much to be desired and so the first half of this chapter will outline a dimension reduction technique, factor analysis. It presents a factor-analytic classification of small company financial ratios which can then be used as a starting point for the prediction of small company failure via conditional probability models.