Endogeneity: How Failure to Correct for it can Cause Wrong Inferences and Some Remedies

Abstract

Although researchers in business and management are becoming increasingly aware of the importance of endogeneity affecting regression analysis, they frequently do not have the right methodological toolkit to adjust for this issue. In this paper we discuss such a toolkit. There are also areas in business and management research which to date seem to be mostly oblivious about the endogeneity issue. We highlight such an area, which studies the question of whether firms that are cross‐listed on a foreign stock exchange are charged premium fees by their auditors. When the same methodology (pooled ordinary least squares) as in the existing literature is used, the existence of an audit fee premium for cross‐listed firms seems to be confirmed. However, once methodologies are used which adjust for the various types of endogeneity (i.e. omitted variable bias, simultaneous and dynamic endogeneity) there is no longer support for the existence of such a generalized premium. Hence, not only do we illustrate that failure to adjust for endogeneity has severe consequences such as drawing the wrong inferences, but we also review various ways to control for the different types of endogeneity.

Does Corporate Culture Affect Bank Risk‐Taking? Evidence from Loan‐Level Data

“…While our results are consistent with the view that the cross-sectional differences across bank culture explain the differences in a bank's willingness to absorb default risk from borrowers, they are still subject to endogeneity concerns (see, e.g., Abdallah, Goergen and O'Sullivan, 2015). While our results are consistent with the view that the cross-sectional differences across bank culture explain the differences in a bank's willingness to absorb default risk from borrowers, they are still subject to endogeneity concerns (see, e.g., Abdallah, Goergen and O'Sullivan, 2015).…”

Corporate Governance and Tobin's Q as a Measure of Organizational Performance

“…The adoption of a generalized least squares estimator is critical in alleviating traditionally important econometric issues, such as potential heteroscedasticity between panels and the autocorrelation within them. However, our model, much like other studies in CG, is spurious to endogeneity (Abdallah, Goergen and O'Sullivan, 2015). Our study is also highly likely to be suffering from endogeneity and, more specifically, from omitted variable bias, since both OP and board composition are jointly determined by unobservable firm-specific variables.…”

Does Board Ethnic Diversity Impact Board Monitoring Outcomes?

“…Identifying causal impacts is likely confounded to some degree by endogeneity problems, associated with unobservable omitted variables and reverse causality (Abdallah, Goergen and O'Sullivan, 2015;Bettis et al, 2014;Ferreira, 2010). The appointment of outside directors is endogenous to firm performance (Adams, Hermalin and Weisbach, 2010;Hermalin and Weisbach, 1988), and minority director appointments could be endogenously caused by firm characteristics which could be invariant or variant with time, observable or unobservable.…”