Desarrollo de un modelo de recuperación en una cartera de crédito castigada

Abstract

The present work presents the construction of a methodology for the evaluation of the behavior of credits, which have gone through a written-off process, after a certain number of days of default. It begins with an analysis of the information provided by a credit recovery company, which is the acquirer of the written-off loans. The starting point is the hypothesis of the feasibility of applying the "Logistic Regression" tool with the stored variables. With this information, the model is generated and the corresponding statistical tests are applied to validate the obtained model. The result of the work is a table of scores that allows us the construction of groups with homogeneous behavior, which will be used by the recovery company to correctly focus their efforts on the collection of their credit operations, minimizing the costs that this task represents.