Developing an Enterprise Risk Management Model through Meta-Analysis and Comparing its Predictive Power with the Hoyt and Liebenberg Model (2011)

Abstract

This study was conducted with the aim of developing an enterprise risk management (ERM) model using a meta-analysis approach and comparing its predictive power with the Hoyt and Liebenberg (2011) model. To achieve this, employing the meta-analysis technique, all Persian-language articles published between 2011 and 2025 were examined. The research is developmental and applied in nature, and an inferential approach was used to interpret the results. The findings of the study were presented in several parts. The first part sought to identify the degree of accuracy of the Hoyt and Liebenberg (2011) model in predicting enterprise risk management. In this section, the results of using the neural network method showed that the mean squared error (MSE) was equal to 0.00080. In the second step, for the development of the enterprise risk management model, a systematic review of the literature demonstrated that a set of quantitative and qualitative variables significantly influence the maturity and effectiveness of risk management processes. These variables include operational growth opportunities, the market-to-book value ratio of equity, the risk-free interest rate, operational efficiency, stock market growth, tax orientations, corporate governance-related indices, corporate social responsibility, the firm’s information environment, CEO efficiency, CEO personality traits, board of directors’ characteristics, audit attributes, and the life cycle. These factors were added to the initial model. In addition, when comparing the predictive power of the developed model with the base model, the results indicated that the base model, after training, achieved an MSE of approximately 0.00080 and an RMSE of ≈0.0283. However, the developed model succeeded in reducing the MSE to 0.000422 and the RMSE to 0.020548. This substantial reduction (over 47% in MSE and more than 27% in RMSE compared to the base model) confirms that adding institutional and operational components enhanced the model’s ability to more accurately reproduce the ERMI index. This indicates that the developed model not only has a lower average error but also improves error stability and significantly reduces the probability of severe errors. Overall, comparing these two structures demonstrates that incorporating new variables from the literature and institutional and financial functions, in addition to emphasizing the core aspects of the base model, significantly improves predictive performance. Based on the findings of this research, it can be concluded that the efforts made in this study resulted in the development of a localized model with superior predictive power compared to conventional enterprise risk management models.