The incremental cash flow predictive ability of accrual models

Prior studies on the incremental predictive ability of accrual models over cash flow models with respect to future cash flows led to conflicting results. This paper extends the model of the accrual process developed by Barth, Cram, and Nelson (2001) by including cash flow implications of growth in future sales. The Barth, Cram, and Nelson model is further modified to allow the incorporation of accrual-based prediction of future sales. This paper also presents an accrual-based cash flow prediction model based on a random walk in cash flows adjusted for the reversal of current payables and receivables. Initial results indicate that this simple model based on the complete reversal of current payables and receivables predicts future cash flows better than models based on current cash flows alone. No initial evidence is found that the more sophisticated accrual-based prediction model developed in this paper and estimated via WLS has incremental predictive power beyond that of the accrual reversal model or the cash flow-based models. However, supplementary analysis using a more powerful estimation procedure where the prior three years of observations are pooled does find that the accrual-based WLS model dominates both the cash flow-based models and the accrual reversal model. Consistent with accruals incorporating predictions of future sales, the paper finds that the accrual-based WLS model (when estimated while pooling the prior three years data) is superior to the cash flow-based model in capturing the effect of future sales on future cash flows. In fact, for 13 of 17 industries, tests cannot detect a decrease in absolute forecast errors when actual sales are substituted for expected sales in the accrual-based prediction model.