Predicting the Relation Between Audit Quality and Share Return Using Artificial Neural Networks

Agency problems arise due to conflicts of interest between managers and shareholders. Auditing is considered an effective solution to limit managers’ power in contractual relationships. This study aims to predict the effect of audit quality on stock return changes. For this purpose, data on companies listed on the Tehran Stock Exchange for the period 2006-2023 were extracted, and a nonlinear prediction approach using artificial neural networks was used to predict stock returns. The outputs obtained from the estimation of the artificial neural networks and the results obtained from the estimation using this method with evaluation scales are (MAE = 0.017, MSE = 0.075, and R=0.986). Concerning the random amount and comparing it with R = 0.986, we find a meaningful relationship between audit quality and share returns. However, such a network had the least error compared to the other networks (MAE=0.027, MSE = 0.004). In this study, using a nonlinear approach, an attempt was made to consider issues related to collinearity between variables and latent factors.