An Empirical Evaluation of an Induced Theory of Financial Ratios

The recent paper of Tippett (1990) explores some analytical properties of accounting ratios on the assumption that the underlying financial aggregates are generated by some form of diffusion process. The present paper uses a long time series of four accounting ratios, taken from the Cambridge/DTI database of the accounts of UK listed companies, to assess whether the models advanced in this paper can be sustained at an empirical level. Of the models tested, only the Logarithmic Random Walk and the Elastic Random Walk, seem to be consistent with the data. Even here, however, there is some evidence that the constant variance (homoscedasticity) assumption may not be satisfied and there is also a suggestion that the normality assumption might not be satisfied by the Elastic Random Walk models. Nonetheless, these models perform as well, if not better, than any of the models previously examined in the literature.