Reexamination of Relationship between Pre-Tax Accounting Income and Cash Effective Tax Rate

Prior studies have attempted to explain the variations of the cash effective tax rate (CETR), a typical measure of tax avoidance, by including various firm specific variables. Recently Edwards et al. (2021) (“EKS”) showed a strong negative relationship between CETR and accounting profit before tax (EBT), suggesting that there is a mechanical relationship between CETR and 1/EBT derived from the linear tax function. and empirically demonstrated that 1/EBT appropriately explains the variation in CETR of U.S. listed companies. In this paper, I applied the EKS model to Korean companies. In appearance, there seem to be a negative relationship between CETR and EBT in Korean companies as in the United States, suggesting the applicability of the EKS model. However, in the further analysis, I find that 1/EBT does not sufficiently explain the variation of CETR of Korea's listed companies. As an alternative to the EKS model, I proposed and empirically confirmed that the seemingly negative relationship between CETR and EBT in my sample firm-years may be due to the effect of Winsorizing CETR at 0 and 1. These results are similar to those of EKS in that control of profit level is necessary when using CETR as a dependent variable. However, they are different from EKS in that the main cause is the Winsorization of CETR, and in such cases, a profit level variable such as ln(EBT) is more appropriate as a control variable than 1/EBT.