Auditors’ Decision Making Process Analyzed Through a Decision Tree Method

• In Event: Chief Audit Executives and Decision Making

Abstract

This study examines the complex relationships between many factors associated with an auditor’s decision to issue a going concern audit opinion. This study utilizes a statistical machine learning method, specifically a decision tree model, that can explore this complex decision process. Using U.S. audit going concern opinion data with 9,087 observations and eighty variables, this study finds that stock market factors are more predictive than client factors or auditor factors, and that stock price and prior-year going concern contained the most relevant information about the current going concern decision. This implies that the stock market has priced in the bankruptcy risk that the auditor independently evaluates when deciding to issue a going concern opinion. This study also implies that audit fees are not related to the going concern decision.

To the best of my knowledge, this study is the first to use U.S. audit archival data and statistical machine learning methods to show multiple interacting factors related to an auditor’s decisions-making process when issuing a going concern opinion. The performance of the decision tree is also found to compare favorably with traditional regression, and the simplicity of decision trees makes them a good approach for practitioners.

Author

• Eo Jin Lee, Florida International University