Mathematics
110 Abdulkader, Abdullah A.,
An empirical investigation of the time-series properties of accounting numbers at the industry level. Florida State University, 1979.*
The purpose of this study was to investigate the time-series properties of quar- terly ccounting numbers at the industry level. The investigation was directed toward the specification of statistical models of a selected industry's sales and earnings numbers, and the evaluation of the predictive ability of these specified models in comparison to previously suggested models. The study utilized Box- Jenkins time-series analysis to specify linear stochastic models for a sample of 74 firms from ten different industries. The research into the time-series proper- ties of accounting numbers has direct implications for the assessment of their information content, their predictive ability, and their utilization in decision making.
The results of this empirical investigation warranted several conclusions about the time-series properties of the sales and earnings numbers the industry level. The Brown and Rozeff (1977) first-order autoregressive seasonal moving average model, adjusted seasonally with a drift, was identified for 14 out of 20 series investigated. However, for forecasting purposes, the Watts (1975), Grif- fin (1977), and Lorek (1979) multiplicative first-order moving average model (011 x 011, in BJ notation) was proven to have high predictive efficiency. Al- though the industry effect on the data analyzed in this study was less than what was anticipated, it should not be regarded as a deterrent to conducting time- series research at the industry or the firm level. Extension of research efforts in- to the time-series properties of accounting numbers is one of the necessary con- ditions for their utilization.
199
ص ٣٥٣