Modelling credit risk : evidence for EMV methodology on portuguese mortgage data

• Autor(a) principal: Borges, Maria Rosa
• Data de Publicação: 2020
• Outros Autores: Machado, Raquel
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.5/20884
• topic: Credit Risk; EMV Models; Mortgage Loans; Default Rates; Vintages

Traditional credit risk models failed during the recent financial crisis and revealed weaknesses in forecasting and stress testing procedures. One of the main reasons for this failure was the fact that they did not include lifecycle and macroeconomic adverse selection effects. The Exogenous-Maturity-Vintage (EMV) models emerged in this context, in the credit risk literature. In this article, we assess the applicability of the EMV models to a dataset consisting of Portuguese mortgage data between 2007 and 2017, to study the determinants of default rates. We obtain and examine the exogenous, maturity and vintage curves from the dataset under analysis, plotting defaults rates through time, under each of the three component’s logic (default rates by calendar period, by age and by vintage). We show that these curves follow the expected behavior. Finally, we identify a set of explanatory variables suitable to be incorporated in an EMV model specification, for forecasting purposes, and discuss the rationality for their inclusion in the model.