In this paper I describe a study of the predictive ability of price-level adjusted and unadjusted historical earnings numbers for a sample of firms in the electric utility industry during the period 1935-1940. The criterion used to assess predictability was an ex post valuation. The ex post valuation was computed by discounting actual net cash flows plus a terminal market value back to base periods which ranged from 1938 to 1941. Ex ante valuations based on adjusted and unadjusted accounting data available at the base periods are compared to the ex post valuations to evaluate the predictive ability of these earnings streams. The test is intended to provide empirical evidence about both the accuracy of predictions based on earnings data in general and the relative predictive abilities of price-level-adjusted and unadjusted earnings numbers. The valuation models used in the empirical test are based on homogeneous risk class models defined by Modigliani and Miller.' The derivation of these models are described in the next section.

Abstract ABSTRACT: In a replacement-cost accounting model, changes in the replacement-cost of assets held during a period are viewed by some as holding gains and losses which are includable in income for the period. The academic literature provides two alternative arguments in support of the holding gains treatment. One argument is that replacement-cost changes represent "cost savings"; the other argument is that replacement-cost changes may be used as "surrogates" for changes in net realizable value or discounted present value. The author of this paper examines the two arguments and concludes that neither is acceptable. He recommends that, if a replacement-cost model is used as a basis for financial reporting, replacement-cost changes be treated as direct adjustments to capital.