Accruals and Forecasting

Sloan (1996), Richardson et al. (2005, 2006) examine how firms' accruals relate to subsequent financial performance. They identify a negative correlation and attribute it to accruals lack of reliability. This paper considers the issue from a different starting point: we forecast sales and expenses separately and argue on prior grounds that accruals are generally informative about the changes/growth in the income statement items. Two accruals variables serve as the primary predictors, year-to-year changes in operating assets and operating liabilities. This framework thus implies 2 forecasting equations and where the RHS of each includes the 2 accrual variables, plus controlling variables. Traditional accounting concepts can be applied to gauge the expected magnitudes of the 2x2 load-factors. Moreover, this framework leads to the hypothesis that the 2 accrual variables have a negative effect on the ROA and earnings forecasts, consistent with the literature. However, a closer look at the estimated load-factors shows some subtleties. First, liability accruals are markedly more informative than asset accruals. Second, while both accrual variables forecast ROA robustly, a shift to earnings weakens the results. Third, the 2 accrual variables are more informative about future performance in case of smaller firms. The empirics also highlight the ways in which financial assets and liabilities influence the forecasting and how their effects differ from those of the (operating) accruals