THE IMPLEMENTATION OF CREDIT RISK SCORECARD MODEL IMPROVE THE ASSESSMENT OF CREDITWORTHINESS IN A PEER-TO-PEER LENDING COMPANY: CASE STUDY AT PT XYZ

• Main Author: Franata/29116117, Rendy
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/30315
• Institution: Institut Teknologi Bandung

PT XYZ is a new entrant in the financial service industry. The company concerns to serve the <br /> <br /> loan installment with sharia scheme. Currently, the company’s name is getting widely known <br /> <br /> and the number of people applying for loans to the company is increasing. It is an opportunities <br /> <br /> as well as challenges for the company to serve well the growth of its customers. The company <br /> <br /> faces a problem in term of credit scoring system, where misinterpretation occurred in its system. <br /> <br /> The misinterpretation increases the possibility of opportunity loss and real loss which the <br /> <br /> company might suffer. The managements have no experience in this field, since the company <br /> <br /> is just been established in recent years. The system is evaluated regularly using trial and error <br /> <br /> method. <br /> <br /> Through this research, the author construct the new model for credit scoring system which is <br /> <br /> based on empirical data using the company’s historical data of approved customers. The author <br /> <br /> chooses the Credit Risk Scorecard Model to predict the customer’s creditworthiness assessment <br /> <br /> output. The model selection is based on the facts that the model has been long successfully <br /> <br /> applied in banking and insurance industry, and also supported by the previous research that <br /> <br /> construct the similar model to predict the default risk. <br /> <br /> The model construction results indicates that the model is better than the company’s existing <br /> <br /> scoring system. It is shown from the Pearson correlation, where the new model has Pearson <br /> <br /> correlation value which is higher significant power than the company’s model. Moreover, the <br /> <br /> new model is tested using cross validation test and results the optimum is occurred in the fitted <br /> <br /> cut-off value of 0.235 with the accuracy rate, sensitivity rate, and misclassification rate, <br /> <br /> respectively as for 69.84%, 77.27% and 28.24%. Moreover, the result is almost similar to the <br /> <br /> previous study. <br /> <br /> Keywords: Credit risk management, credit scorecard, creditworthiness, logistic regression, <br /> <br /> p2p lending.