Omitted Variable Bias in Regression Analysis Internal Vailidity Assessment of Empirical Models in Capital Markets-Based Accounting Research

DOI: 10.5176/2251-1997_AF18.223

Authors: Kazuhiro Manabe

Abstract:

This study examines the effect of omitting earnings change in regression models derived by Easton and Harris (1991). The sample selected is from listed non-financial firms on the first section of the TSE for the period 2001–2016. Omitting earnings change in regression models could cause an increase in the parameter of earnings level. This increase might be due to upward bias caused by omitting earnings change and not due to the multicollinearity of the earnings variables.

Keywords: value relevance research; omitted variable bias; internal validity; capital markets-based accounting research