Construção de modelos de classificação de risco de crédito para empresas brasileiras com base em indicadores contábeis / Construction of credit risk classification models for Brazilian companies based on financial ratios

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

This study demonstrates, through the utilization of discriminant analysis and logistic regression techniques, the predicted power of credit risk models based financial ratios. The sample utilized is composed of 126 Brazilian companies negotiated in BOVESPA, of which 63 are solvent and 63 are insolvent, for the period between 1988 and 2006. The models were estimated for the entire period, for the period before the implementation of Real Plan and for the period after the Real Plan. It was observed a significant improvement of the model ofter the implementation of the Real Plan. Despite the rigorous assumptions, the discriminant analysis model had a great discrimination between solvent and insolvent companies, and the level of adequacy was superior to the logistic regression model. It was observed that the utilization of financial ratios in credit risk classification models permits the discrimination of companies in default and companies not in default, with a level of adequacy above 90%. Three main questions were covered: (i)the evolution of financial ratios - a comparison of the importance of different financial ratios before and after The Real; (ii) determination of a credit risk classification model; and (iii) which statistical technique has greater prediction power of default: discriminant analysis or logistic regression

ASSUNTO(S)

modelo de escoragem risco de crédito discriminante analysis bankruptcy forecast análise discriminante previsão de falência model of escoragem credit risk countable pointers economia regressão logística indicadores contábeis logistic regression

ACESSO AO ARTIGO

Documentos Relacionados