Nonlinear Regression Model Based on Fractional Bee Colony Algorithm for Loan Time Series

Abstract

High levels of nonperforming loans provide negative impacts on the growth rate of gross domestic product. Therefore, predicting the occurrence of nonperforming loans is a vital issue for the financial sector and governments. In this paper, an intelligent nonlinear model is proposed for describing the behavior of nonperforming loans. In order to find the optimal parameters of the model, a new fractional bee colony algorithm (BCA) based on fractional calculus techniques is proposed. The inputs of the nonlinear model are the loan type, approved amount, refund amount, and economic realm. The output of the regression model is that whether the current information is for a nonperforming loan or not. Consequently, the model is modified to detect the status of a loan. So, the modified model predicts the occurrence of a nonperforming loan and determines the loan status, i.e., current, overdue, and nonperforming. The proposed procedure is applied to data gathered from an economic institution in Iran. The findings of this study are helpful for the managers of banks, and financial sectors to forecast the future of the loans and, therefore, manage the budget for the upcoming loan requests.