Resumen
AbstractFailure prognosis for companies with due allowance for time dimensions In the United States of America it was initially attempted to predict financial failure by means of compiling a failure-prognosis model consisting of a number of weighted ratios. The model developed was a general model which only examined one time dimension of failure, namely, differentiating between solvent and insolvent companies one year prior to financial failure. The model was applied, without any adjustments, in penultimate and prior years, without ascertaining whether the change in time dimension has an influence on the classification results of the model. A similar model was later developed for South African circumstances. This model had the same shortcomings. A possible problem, which was quite likely not taken into account in the compilation of the existing models, was whether a general failure-prognosis model provided the best classification results in all possible circumstances. In this I article endeavour to determine whether or not failure-prognosis models, which are compiled for specific circumstances, provide more accurate classification results than a general model. Failure-prognosis models were compiled separately for companies, for periods of one and four years preceding financial failure. The models were interchanged and it was found that results were unreliable when the incorrect model was applied. Models compiled to predict business failure within the first year, for example, did not fare well when applied to earlier years and vice versa.