On the Relationship between Accruals and Cash Flows: Earnings as a Confounding Variable

• In Event: The Relation Between Earnings, Cash Flows, and Accruals

Abstract

Bushman, Lerman, and Zhang (2016) present remarkable evidence contradicting the long-held notion that accruals and cash flows are negatively correlated, reporting that the overall correlation and explanatory power have largely disappeared in recent years. We postulate that the absence of a consistent negative correlation is due the lack of consideration that there exists a “confounding” variable (Elwert and Winship 2014): earnings. We demonstrate that, once we control for earnings by conditioning the accrual–cash-flow regressions based on earnings on a portfolio basis, the negative correlation and explanatory power resurface in a strong and consistent manner. We present a framework that helps understand the accrual–cash-flow relationship in the context of basic accounting income recognition principles. This framework also helps connect the phenomenon of increasing earnings volatility, and the correlation between earnings and cash flows, to deteriorating explanatory power of the accrual–cash-flow models over the past five decades.

Authors

• Sok-Hyon Kang, George Washington University

• Hyun Jong Na, George Washington University

• Seunghee Yang, Seoul National University