An empirical investigation of time series properties of quarterly cash flows and usefulness of earnings in predicting cash flows

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Business Administration


Badr Ismail


univariate models bivariate models

Subject Categories



An accurate prediction of the firm's future cash flows requires the knowledge of the underlying cash flow generating process. Accordingly, this study has examined the time series properties of firms' quarterly net cash flows cross-sectionally and derived two single models (ARIMA (000)(100) and ARIMA (000)(011)) that can be assumed to have generated the series. The results indicate that the time series properties of cash flows are different than those of earnings.

Another issue addressed in this study is the role of earnings in predicting future cash flows. While there is general agreement that firm value is related to future cash flows and thus an accurate prediction of cash flows is of great importance, there is controversy about the usefulness of earnings in predicting future cash flows. The issue is tested by comparing the predictive accuracy of univariate (cash flows) models with that of bivariate (cash flows and earnings) models for each firm. If there is information on cash flow series contained in the historical data of earnings, more accurate cash flow forecasts should result when the two series are modelled together. A multivariate state space method was used to obtain the firm-specific univariate and bivariate models. The results indicate that bivariate models provide statistically more accurate cash flow forecasts than the univariate models, or equivalently, earnings convey information about future cash flows that is not conveyed by past cash flows. The findings of this study provide strong support for the accrual system in the context of cash flow prediction.

Finally, to determine the model that produces the most accurate cash flow forecasts, the predictive accuracy of firm-specific univariate and bivariate state space models were compared with the predictive accuracy of the two single models identified from the cross-sectional analysis. Based upon the one-step-ahead forecasting performance of the models, it was concluded that the ARIMA (000)(011) model performed very well vis-a-vis other models. Consequently, the ARIMA (000)(011) model is suggested to be a good description of the stochastic process generating quarterly cash flows in general, and for individual firms.


Surface provides description only. Full text is available to ProQuest subscribers. Ask your Librarian for assistance.