A influência da localização geográfica na inadimplência das empresas atendidas por um banco de varejo utilizando a regressão logística geograficamente ponderada

Abstract

This dissertation proposed the use of the methodology of Geographically Weighted Logistic Regression (GWLR) for the construction of credit scoring models. This technique uses the geographical location of credit borrowers to consider observations in the development of the model for each region of the study. We used real data referring to working capital operations granted by a financial institution the Micro and Small Enterprises of the State of Paraná. The results show that with the use of the GWR model, financial institutions have been able to identify in which regions there is a greater or lesser influence of the variable default and with that provide punctual corrections in regional credit managers and / or control the credit supply in a given region . It is concluded that the use of this technique by financial institutions is feasible and can bring significant contributions to reduce delinquency in financial institutions and can be applied in the development of credit scoring models.