Identifying Informative Audit Quality Indicators (IAQI) Using Machine Learning

Researchers have tested a wide array of theories and developed a broad set of explanatory variables for audit quality. However, little is known about which of these audit-related variables are the most predictive of audit failure (i.e., low-quality audits). We devote this study to provide researchers and regulators with a portfolio of informative audit quality indicators (IAQI) - publicly available theory-driven audit-related variables that can best predict audit failure. We adopt machine learning, a computational method that can navigate through a long list of variables and identify a subset that can make the best out-of-sample predictions. By applying various machine learning algorithms, we identify 11 audit-related variables as IAQI with their predictive power validated. These IAQI reflect auditor competence, independence, effort, incentive, and the quality of the audited financial reports. Our study provides researchers, regulators, audit firms, and investors with a list of IAQI that are predictive of audit failure