Predicting the probabilities of default for the banking sector in the United States

• Autor(a) principal: Weisel, Lukas Klaus
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Texto Completo: http://hdl.handle.net/10400.14/29819
• topic: Structural credit risk models; Banking; Probability of default; Modelos de risco de crédito estrutural; Banca; Probabilidade de inadimplência

This dissertation implements a structural credit risk model to predict the probabilities of default for the eight largest retail banks in the United States. Similar to Goldstein, Ju, and Leland (2001), the model implemented in this dissertation relates the project/asset value with the firm capacity to generate earnings. Most papers in the literature consider that firms have fixed financial costs and that shareholders decide whenever to liquidate the firm based on the distance between some underlying earnings measure and these fixed costs. This assumption is not reasonable in the case of banks. As an alternative to incorporate shareholders strategic default decision, non-financial fixed costs are considered. These non-financial fixed costs are defined as the non-interest expenses and proxy the operational leverage of the bank. The analysed sample period contains 19 consecutive years, covering the dotcom crisis, the financial crisis and several minor crises in between. The model was calibrated by applying the iterative approach proposed by Vassalou and Xing (2004). The average computed probability of default for the whole sector ranged between 0.06% in 2006 and 5.80% during the financial crisis. These results were compared with the probabilities of default and the distances-to-default implied by Moody’s and Standard & Poor’s credit ratings for the period between 2006 and 2018. Though the probabilities of default show a low not significant correlation, the distances-to-default have a correlation of 53.27%, which was found to be significant at the usual significance levels.