A Comparative Study of the Time-Series Behavior and Predictive Ability of Historical Cost Earnings and General Purchasing Power Adjusted Earnings: An Empirical Investigation

Shalchi, Hossein

Permalink

https://hdl.handle.net/2142/71361 Copy

Description

Title

A Comparative Study of the Time-Series Behavior and Predictive Ability of Historical Cost Earnings and General Purchasing Power Adjusted Earnings: An Empirical Investigation

Author(s)

Shalchi, Hossein

Issue Date

1981

Department of Study

Accountancy

Discipline

Accountancy

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

Business Administration, Accounting

Abstract

• The knowledge concerning the time-series properties and predictive ability of accounting numbers are of considerable importance for various areas of research. However, there exists no descriptive evidence in relation to the underlying time-series behavior and predictive ability of general purchasing power (GPP) accounting numbers. Accordingly, the primary objective of this study is to gather empirical evidence on the time-series behavior of both historical cost and GPP adjusted earnings series, and to use the Box-Jenkins (B-J) methodology to compare the relative predictive ability of the historical earnings series against the GPP annual earnings series. Both deflated and nondeflated earnings were examined. Also, in order to assess the effect of the monetary gains or losses on the time-series behavior of GPP data, each of the income measures were calculated with and without inclusion of the monetary gains or losses.
• A total of thirty years of financial statement data were collected for each of the 47 firms included in the final sample. The historical financial statements were restated by using a procedure in conformity with FASB Statement No. 33. For each of the 47 firms in the sample, the deflated and nondeflated earnings series for the period 1952-1977 were used to determine the appropriate B-J firm-specific time-series models. These moels were utilized to generate one-step-ahead, and two-step-ahead forecasts. Then the base period was updated to include 1978 data, and the appropriate time-series models were identified and estimated. These models were then used to generate point forecasts of the same earnings measures for the year 1979.
• The results indicated that the time-series behavior of the majority of the series were suggestive of either an autoregressive or a white noise process. Also, the results indicated that the time-series behavior of the earnings measures were different across the alternative income accounting measurement methods. The results provided no support for the random walk model as a good representative of the underlying process generating earnings series. This result has very important implications for various areas of research both in theoretical and empirical contexts.
• The second purpose of this research was to compare empirically the relative predictive ability of the alternative accounting earnings to predict their own future values. Two measures of prediction accuracy were employed. Statistical tests were performed to test the hypothesis of no difference between historical cost and the two general price-level adjusted earnings forecast accuracy. The results of the predictive ability analysis indicated that no single accounting method outperformed its counterparts in all cases. However, based on the statistical results, it was concluded that the forecasts based on general price-level adjusted data with inclusion of monetary gains or losses were generally inferior to the predictions based on the other earnings measures. This finding indicates that the exclusion of the monetary gains or losses in the computation of net income (as recommended by FASB Statement No. 33) appears to be preferable to their inclusion.

Graduation Semester

1981

Type of Resource

text

Permalink

http://hdl.handle.net/2142/71361

Owning Collections