Selection Bias in Audit Firm Tenure Research

Abstract

Prior studies find that long auditor tenure improves audit quality. An alternative explanation, however, is that low audit quality increases auditor turnover, thereby self-selecting into the short-tenure group. To disentangle these explanations, this paper examines selection bias in audit firm tenure research in three steps. First, we empirically demonstrate the existence of self-selection. We find that the length (or duration) of the auditor-client relationship is not random: firms that have higher audit quality, higher profitability, and lower risk are more likely to have long relationships with their auditors. Second, we use simulated panel data to illustrate that, when self-selection exists, results consistent with long tenure improving audit quality may be obtained, even if lengthened tenure has no impact or even an adverse impact on audit quality. We show that incorporating fixed effects is critical to address the selection bias. Third, we examine commonly used audit-quality proxies and find that, after controlling for selection bias, there is little evidence that auditor tenure affects audit quality or perceived audit quality. Our results suggest that the previously documented positive association between auditor tenure and audit quality reflects selection bias, instead of a treatment effect.