Reflexo das flutuaÃÃes macroeconÃmicas para a qualidade do crÃdito concedido a pessoas jurÃdicas: estudo de generalizaÃÃo de redes neurais
AUTOR(ES)
ClÃudio Freitas GonÃalves
DATA DE PUBLICAÇÃO
2002
RESUMO
This study goes over the adequateness of the usage of a special class of adaptive systems â the artificial neural networks â in the sampling of the payment default due to macro-economical instability. The negative balance for seven different categories of loans granted to corporate bodies are sampled through neural networks. The obtained valuations are compared to those observed in the application of a traditional econometric model, based on multiple regression. The present work begins exploring the Brazilian Economical History of the past 50 years in order to understand the development of credit in this period and its connections with the macro-economical environment. In order to admit the approach of the used methodology, a revision of the literature concerning artificial neural networks is made, pointing both advantages and disadvantages of its employment when compared to traditional econometric methods. This work also carries out an empiric research of the presented methodologies and understands that the usage of neural networks is practicable. The usage of such networks presents, in most of the studied cases, finer outcomes than the ones obtained with the model of linear regression in the sampling of the overdue credit
ASSUNTO(S)
economia macroeconomic scenarios neural networks risco de crÃdito cenÃrios macroeconÃmicos credit risk redes neurais
Documentos Relacionados
- InvestigaÃÃo sobre o efeito de ruÃdo na generalizaÃÃo de redes neurais sem peso em problemas de classificaÃÃo binÃria
- AplicaÃÃo da GeneralizaÃÃo CartogrÃfica em Mundos Virtuais
- Liberdade e decisÃo judicial : a autoridade da vontade como instrumento legitimador das razÃes jurÃdicas
- Da improbidade administrativa : formalizaÃÃo e generalizaÃÃo no direito ambiental
- EficiÃncia do mercado acionÃrio brasileiro : retorno das aÃÃes negociadas na Bovespa, variÃveis macroeconÃmicas, causalidade e fatores condicionantes