Статистически модели за управление на кредитния риск. Сравнение на логистична регресия и анализ на оцеляването
Statistical Models for Credit Risk Management. Comparison Between Logit And Survival Analysis

Summary/Abstract: The main purpose of the article is the development and implementation of two main scoring models for segmentation and effective risk management of non-performing loans, namely: logit and survival analysis. In the current economic situation, the number of non-performing loans is growing rapidly, which implies a higher level of expenditures from collection activities and underscores the need for a model optimizing this process as a crucial part of credit risk management. The most popular statistical model in this field is the logistic regression. The goal of statistical modelling is to segment the nonperforming portfolio and to apply different collection strategies to the different segments. The strategies can be applied based on the estimated probability of early repayment of the overdue debt, the probability of long-term insolvency and so on. This peculiarity suggests the need of additional dimension offered by the survival analysis, supplying the ability to predict not only„if“, but „when“ an observed event will occur.

• Details
• Contents