Análise de sobrevivência aplicada ao tempo de inadimplência de clientes de uma empresa de crédito

Abstract

With the increase in the number of companies offering credit in the country and the large volume of products offered, which reach the most diverse customer profiles, it is necessary for institutions to effectively control risk, in order to guarantee the improvement of credit indices. default, fraud and collection, which are essential variables for the solvency of a financial institution. Therefore, in addition to a credit scoring model that takes into account the institution’s credit policy and risk appetite, the analysis of clients that are already part of the portfolio through named behavior scoring models has been gaining more and more space, as they are able to reassess the customer and guide the company’s short-term decisions. Among the various techniques used to build these models, survival analysis has been gaining ground for being able to analyze the time until the event of interest. In this context, the objective of this work is to apply survival analysis techniques in the construction of a behavioral model, for 94,407 customers of a multinational retail company that offers the credit card product, understanding the relationship of the selected variables in the default process . The selected parametric model presented adjustment for all selected covariates, allowing the identification of some characteristics that influence the time until default of the observed customers.