Sensitivity Analysis of Banks Efficiency to the Financial Variables

Abstract

Discriminant analysis is a classification method that can predict the group membership of a newly sampled observation. In discriminant analysis, classification of new observed data has an uncertainty. In this paper, the confidence degree is introduced to determine the confidence of classification of new observed data. Then, a Monte Carlo-based sensitivity analysis is applied to an assessment of the influence of financial variables on the overall variance in defined confidence degree. The variables that have a dominant influence on the confidence of new observed data are identified. The result show that in changing classification of the new branch of banks, the uncertainties in the long-term deposits and loans granted variables are more significant than the uncertainties in other bank's variables in decision making. Also, the personnel, non-benefit deposit and non-performing loans variables have little effect. The results open the door to better making decision on establishing a new bank's branches.