A non-default fraction bivariate regression model for credit scoring: An application to Brazilian customer data

• Autor(a) principal: Cancho, Vicente G.
• Data de Publicação: 2016
• Outros Autores: Suzuki, Adriano K., Barriga, Gladys D. C. [UNESP], Louzada, Francisco
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1080/23737484.2016.1243456; http://hdl.handle.net/11449/220953
• topic: Bivariate non-default rate models; competing risks; cured fraction; long-term survival models

In this article, we propose a lifetime model for bivariate survival data with a non-default rate. Our approach enables different underlying activation mechanisms that lead to the event of interest. A number of competing causes which may be responsible for the occurrence of the event of interest are assumed to follow a Poisson distributions, and a positive stable distribution was considered for the frailty component. The Markov chain Monte Carlo (MCMC) method is used in Bayesian inference approach and some Bayesian criteria are used for a comparison. Moreover, we conduct the influence diagnostic through the diagnostic measures in order to detect possible influential or extreme observations that can cause distortions on the results of the analysis. The proposed models are applied to analyze a Brazilian customer data set.