Modelo predictivo del comportamiento crediticio considerando el rendimiento académico de los estudiantes de la Universidad Técnica Particular de Loja durante los años 2022 y 2023.

Abstract

Abstract: This research aims to evaluate the payment behavior of students with educational loans at the Universidad Técnica Particular de Loja (UTPL), in the distance and online learning modality, during the 2022-2023 academic periods. Using a Logit model, academic, financial, and sociodemographic variables were considered. The descriptive analysis revealed that 78% of the students with loans are continuing students, 59.4% are female, and the predominant age range is 26-35 years. The Logit model achieved an area under the curve (AUC[LM1.1]) of 0.81 and a pseudo-R² of 0.26. The odds ratio (OR) for outstanding balance (OR = 3.42), age (OR = 1.31), female gender (OR = 0.92), geographic area, and academic performance (β = -0.53[LM2.1]) significantly influence payment behavior. Students with a GPA below 7 points have a 69.81% default rate, confirming that academic performance can be used as an early warning system for potential loan defaults. These findings will allow for the design of early warning systems and student support strategies that integrate the academic and financial departments of the institution.