The Confounding Effects of Cost Stickiness on Conservatism Estimates: A Replication Study of Banker et al. (2016) Using Japanese Data

Abstract

Standard conservatism estimation models quantify the magnitude of conservatism by regressing earnings on stock returns. However, Banker et al. (2016) point out that cost stickiness may be a confounding factor in conservatism estimation. Cost stickiness makes earnings behavior asymmetric when sales increase and decrease, resulting in an asymmetric relationship between earnings and stock returns. Our analysis provides evidence that conservatism estimates are overestimated by 20.7% in Japan if we do not control for cost stickiness on conservatism estimates. We also conducted a replication test of Shuto and Takada (2010), who examined the empirical relationship between managerial ownership and conservatism, and found that their results are partially less robust when they control for cost stickiness. Our results indicate that cost stickiness is also a confounding factor in studies of conservatism using Japanese data.