Aplicação do modelo ZAIG (Zero Adjusted Inverse Gaussian) na análise de uma carteira de cartões de crédito / Application of ZAIG model (Zero Adjusted Inverse Gaussian) in the analysis of a credit card portfolio

AUTOR(ES)

Rafael Rodrigues Troiani

DATA DE PUBLICAÇÃO

2009

RESUMO

The ZAIG (Zero Adjusted Inverse Gaussian) model is based on a semicontinuous distribution with positive asymmetry and with concentration in 0. This paper applies ZAIG model in a credit card portfolio, analyses its applicability as a behavior score and evaluates its quality in predicting the financial loss of the portfolio. As a behavior score, the ZAIG model presented results that indicate good performance, especially to discriminate the loss value of each observation, instead of the probability of default. As a portfolio loss prediction model, ZAIG presented results close to the observed and slightly superior than the naïve model. Applied in a sample from another period of time, the model still discriminates well each observation, but it loses accuracy in predicting portfolio loss

ASSUNTO(S)

risco de crédito credit risk zaig zaig economia cartão de crédito credit card

Documentos Relacionados

• Aplicação do Modelo Multi-estado de Markov em Cartões de Crédito
• A comparison of classification methods applied to credit card fraud detection
• Estimating total claim size in the auto insurance industry: a comparison between tweedie and zero-adjusted inverse gaussian distribution
• Modelo preditivo para perda de crédito e sua aplicação em decisão de spread
• PRODUCT DEVELOPMENT TO CREDIT CARDS INDUSTRY: A CONJOINT ANALYSIS APPLICATION