Developing of NPA predictive Model for Pawning Advances in Sri Lankan Banking Industry

Abstract

The banking sector in Sri Lanka has been a major contributor to its profitability since the regulation of pawning in 1942 and its commencement in banks in 1961. The business of pawning has been influenced by global economic events such as the 2008 financial crisis and the subsequent increase in gold prices. However, with falling global inflation and a decrease in gold prices in 2013, the risk of nonperforming loans in the pawning and gold-backed loan segments increased for Sri Lankan banks. Four models were developed to evaluate the performance of the pawning business and the best models were found to be M1 (PCA with six features and SVM) and M3 (RFECV feature elimination with Logistic Regression), both with accuracy scores of 95% and 96% respectively. It was observed that RFECV selected all 10 features in the dataset.