Firms which elect to use accelerated depreciation for tax purposes while using straight-line for reporting purposes have only one approved method for recording the difference between book tax expense and the tax liability.' This method relies on a deferred credit account in which the difference between the tax expense (calculated using straight-line depreciation) and the actual tax paid is recorded. The reported net income resulting from this procedure is the same as it would have been if straight-line depreciation were taken for tax purposes. Tax allocation is considered an extension of accrual accounting to include` taxes. The logical grounds for such inclusion are laid in the classification of taxes as expense items.

One criterion being employed by a growing body of empirical research is predictive ability. According to this criterion, alternative accounting measurements are evaluated in terms of their ability to predict events of interest to decision-makers. The measure with the greatest predictive power with respect to a given event is considered to be the best method for that particular purpose. Because the predictive ability criterion is currently being used and is likely to experience even greater use in the future, this article examines its origin, its relation-ship to the facilitation of decision-making, and the potential difficulties associated with its implementation. Knowing the origin of the predictive ability criterion is important in understanding what is meant by predictive ability and why it is being used in evaluating accounting data. The criterion is well established in the social and natural sciences as a method for choosing among competing hypotheses. The use of the predictive ability criterion presupposes that alternatives under consideration have met the tests of logic and that each has a theory supporting it.