A Machine Learning Approach to Assessing Audit Quality in Company with Non-Switching Auditors: Random Forest Classifier Model

For many years, legislators have been concerned that the change in the auditor, along with the incentive to buy the audit opinion, can have a negative impact on the quality of the audit. Therefore, in all the researches conducted in this field, the audit quality after the change of auditor has been investigated. This study investigates the effect of auditor change probability on audit quality using a Random Forest Classifier model. In this research, using the machine learning technique and Random Forest Classifier model, the probability of auditor change in companies listed on the Tehran Stock Exchange for the years 2003 to 2021 and the effect of this probability on audit quality in companies without a change in auditors have been reviewed. The results show that the companies in which the probability of auditor change is high; They have a lower audit quality. In the following, according to the hypotheses related to the reduction of audit costs in large companies based on the familiarity discount framework, the above result has been analysed separately in large and small companies. The results show that larger companies, where there is a possibility of changing auditors, experience a greater decrease in audit quality. Such results indicate that, by using the model presented in this research, legislators and investors can identify the behaviours that occur by auditors and under the pressure of audited companies to obtain desirable results.