EN
One of the central tasks of credit institutions is credit risk assessment, in which the estimation of the probability of default is an important element. The size of an institution’s credit portfolio can decrease as a result of early repayments, which changes the probability of default over time. Prognosis of the probability of default should therefore also take into consideration the prognosis of early repayments. In this paper, methods of evaluating the probability of default over time, using competing risks regression models, are considered. Methods of evaluation for models of default over time are proposed. A sample of retail credits, provided by a Polish financial institution, was empirically examined.