Modelagem da perda esperada: uma alternativa para tratar o efeito da correlação entre a PD e LGD

• Autor(a) principal: Pazzoto, Bruno Bortoluzzo
• Data de Publicação: 2012
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional do FGV (FGV Repositório Digital)
• Texto Completo: http://hdl.handle.net/10438/10073
• topic: Credit Risk; Expected loss; Risco de crédito; Perda esperada; Economia; Administração de crédito; Administração de risco; Comitê de Supervisão Bancária da Basiléia

With the relevance of the credit market has been gaining in the economy this study set out to do a conceptual review of credit risk. Since the expected loss as the main component of credit risk, the work proposes a new way to calculate it. The way it is modeled usually presuppose that the input parameters PD and LGD are independent. Some authors have questioned this assumption and that if this dependence is not taken into account calculations of expected loss and capital should be allocated will be incorrect. An alternative to treat the correlation is modeling the two components together, comparing the results of the usual model with the new model the conclusion is that the error of estimate of expected loss with the second model has been smaller. We can not say that the smallest error in the estimate of loss is because of the correlation between PD and LGD, but modeling the parameters together retires this strong assumption.