ASSESSMENT OF THE CREDITWORTHINESS (CREDIT RATING) AND CREDIT RISK OF THE POTENTIAL LOAN APPLICANTS – COMPANIES IN THE REPUBLIC OF SRPSKA, USING LINEAR DISCRIMINATORY FUNCTIONS

Abstract

One of the scientifically proven and effective methods for managing credit risk is a credit rating system based on client?s solvency. The model for the credit rating of bank customers presented in this paper is based on Fisher linear discriminatory analysis (FLDA), primarily due to its robustness and ease of use. As a rating tool proposed FLDA can advance the process of customer screening and rating based on the corresponding pairs of input variables where the relatively high security generate a clear distinction between ?good? and ?bad? customers. The paper is based on 2009 year end data from reports of financial results of companies from the Republic of Srpska, submitted to the Agency for mediation, information and financial services (APIF RS). As a control factor data on the history of servicing loan obligations to commercial banks (for the same companies), contained in the reports of the Central credit registry of legal entities and natural persons of BiH (BiH CRK) was used. The results suggests that pairs of indicators: 1 return on assets (ROA) and return on equity (ROE); 2 total income of 000 KM and creditworthiness, and 3 credit rating and liquidity have accurate predictions over 80%, which is statistically significant probability, and as such can be used for credit classification of future bank clients.