Análise da mortalidade de empresas: uma abordagem com análise de sobrevivência

Abstract

Credit granting involves different trajectories over time, as contracts can be paid off, become delinquent, or remain under monitoring without an event occurring until the end of the observed period. In this context, this work aimed to apply Survival Analysis techniques in a competitive risk scenario, considering the outcomes of default and repayment in a simulated portfolio of renegotiated credit contracts. To this end, a fictitious database of 10,000 contracts monitored for 12 months was constructed, containing registration, financial, and behavioral variables of the companies. Initially, descriptive analyses were performed to characterize the contracts and the observed outcomes. Subsequently, the Cumulative Incidence Function was estimated for each event of interest and compared with the Kaplan-Meier estimator. Cause-specific Cox models and Fine-Gray models were also adjusted to assess the association of covariates with the analyzed outcomes. The results indicated that default was the most frequent event and occurred, on average, earlier than repayment. The comparison between Kaplan-Meier and the Cumulative Incidence Function reinforced the importance of using specific methods for competing risks, especially because treating the competing event as censoring can overestimate the cumulative probability of the event of interest. In Fine-Gray models, the cumulative incidence of default was associated with greater financial fragility and worse payment behavior, while repayment was related to a better credit profile, greater stability, and a higher percentage of payments made on time. It is concluded that Survival Analysis with competing risks constitutes a useful approach for studying the temporal dynamics of credit portfolios, allowing for the adequate differentiation of mutually exclusive events and the incorporation of censored observations.