Methodological Issues with Common Accounting Variables in the Context of Measuring Income Shifting Consequences
51 Pages Posted: 25 Jun 2020
Date Written: June 2, 2020
We document the empirical limitations of two common types of measures used in accounting research: additive linear scores and first-stage coefficients. While they have valid applications, using these measures in hypothesis testing can lead to misleading inferences. We describe these measures and their limitations, uses, and misuses in the context of recent literature that examines the consequences of income shifting. After illustrating why these measures are not ideal for empirical hypothesis testing, including examining income shifting consequences, we develop and validate a new measure of income shifting that can move the income shifting literature forward. This measure can be parsed to separately identify exogenous and endogenous variation in income shifting, and to separately analyze the U.S.–foreign and foreign–foreign components of income shifting.
Keywords: Income shifting, accounting measurement, correlated omitted variables
JEL Classification: C18, C81, F38, H26, H32, M40
Suggested Citation: Suggested Citation