Revisiting the Common Assumptions About the Desirability of Earnings Smoothness

Abstract

Using a sample of all firm-year observations for the period 1990 to 2010, this study investigates (1) the theoretical assumption that earnings smoothness enhances earnings predictability through increasing earnings persistence, (2) informativeness of high-smoothed earnings versus low-smoothed earnings, and (3) the conventional wisdom that highly smoothed earnings have less risk premium, thus, have lower cost of capital. Robust empirical models are developed to address the three research questions. We find that in contrast to the common assumptions, earnings smoothness improves forecast accuracy but not through increasing the persistence of earnings. The findings support the informativeness hypothesis that earnings announcements of lower-smoothing firms convey more value-relevant information than higher-smoothing firms. Finally, the study supports the third hypothesis that earnings smoothness leads to lower cost of capital through certain accounting practices. Contrary to the common belief, the results suggest that earnings smoothness is not a desirable property and caution needs to be exerted before engaging in more earnings smoothness.