A High-Dimensional Approach to Predicting Audit Opinions

ABSTRACT

This study develops a model for the prediction of audit reports. The research data comprises 57881 firm-year observations for public companies listed on the New York Stock Exchange (NYSE), the American Stock Exchange (AMEX), and the NASDAQ from 2000 to 2019. The dataset consists of a high dimension of predictor variables (105 variables), including accounting-based, ownership concentration, executive compensation, market price, analysts rating, macroeconomic, and audit-related variables. A commercial version of Gradient Boosting, called TreeNet®, is used to build the prediction model. The results indicate that the developed model demonstrates high performance in predicting going-concern reports with an accuracy rate of 97.5%.

KEYWORDS:

• Audit opinion prediction
• machine learning
• auditing
• going-concern opinion
• gradient boosting

JEL CLASSIFICATION:

• M42
• M41
• M21

Acknowledgments

I would like to thank the editor and anonymous reviewers for their invaluable comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ fred.stlouisfed.org.

Additional information

Funding

The work was supported by the Faculty Research Grant, Vice Chancellor for Academic Affairs, University of Minnesota Crookston [1000-10670-20089, 2020].