The Earnings Smoothing Potential of Systematic Depreciation

Ample evidence exists that managers attempt to use ‘artificial’ choices concerning alternative accounting procedures for classification, valuation and allocation of transactions to smooth reported periodic income or earnings series. However, the actual smoothing effect of such choices and their influence on the properties of earnings numbers is unknown. This study analyses the smoothing potential of depreciation when systematically applied. It exploits an idea regarding the effects of inclusion of expectations information in earnings calculations due to Willett (1988, 1991b), theoretically elaborated by Gibbins and Willett (1997) and applied analytically by Lane and Willett (1997). Using a statistical activity cost theory (SACT) framework, we quantify the smoothing (variance reduction) potential of straight‐line and reducing‐balance depreciation methods and compare the potential with an analytically derived depreciation method which optimizes variance reduction. We also evaluate the effects of conditioning factors such as accuracy of managers’ expectations and relative asset acquisition and disposal values upon smoothing potential. We find that systematically depreciating assets can smooth undepreciated earnings numbers. Straight‐line depreciation can exploit virtually all of this potential and reducing‐balance depreciation a substantial proportion. However, realising this potential varies greatly depending on conditioning factors. The investigation is practical and policy oriented but complementary to theoretical extensions of the sort suggested by Butler et al. (1994). While the article focuses on depreciation, the SACT framework is completely general and may be extended to any choice of method for smoothing earnings series.