Assessing artificial neural networks (ANNS) adequacy against econometric models for decision making approaches in banking industry

The purpose of this paper is to test the effect of non-parametric methodology of ANNs on enhancing decision-making procedures compared to classic multivariate Regression models. We implement the two methods on decision-making for loan allowances and on a valuation of collaterals using data from a Small Medium Greek Bank. Using sensitivity analysis, we also find the key points where decision-making results change. We provide all the estimations and a comparable matrix is produced in order to point out the similarities and differences between the two methodologies. Also, the key points in decision-making approach are presented. The practical implication is that crucial for decision-making approach in the banking industry and affects bank's risk management. From the presentation of comparison between the two methodologies, we also provide the relationship between independent variables along with decision-making options. The debate in academic society for theoretical and non-theoretical approaches in decision-making procedure is crucial the last decades. Our paper contributes also to this academic debate, by presenting advantages and disadvantages per methodology. This study demonstrates a practical tool for decision roles and units who are trying to define the best practice for decision-making procedures.