ABSTRACT
This chapter describes Univariate models perform on British data. It should then be possible to compare their relative forecasting ability against that of a number of multivariate models. The six continuous qualitative variables studied were: the reporting lag between financial year end and publication of the accounts; the log of a company's turnover; and year-on-year changes in the reporting lag, in auditors' remuneration, in directors' remuneration, and in directors' shareholdings. Logit models were used to assess the discriminatory power of a number of individual financial ratios. These measured various company attributes - namely profitability, liquidity, gearing, size and asset turnover. When the intercept term was suppressed, only profitability seemed to have much explanatory power in distinguishing between failing and non-failing firms. The major problem with all these univariate models, however, is that if allowance is made both for sampling bias and for the costs to decision makers of misclassifying surviving companies as bankrupt, the operational usefulness of the models is greatly reduced.
