Polynomial distributed lag model for stress testing Brazilian financial system

• Autor(a) principal: Zaniboni, Natália Cordeiro
• Data de Publicação: 2019
• Outros Autores: Montini, Alessandra de Ávila
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Revista de Ciências da Administração
• Texto Completo: https://periodicos.ufsc.br/index.php/adm/article/view/2175-8077.2019V21n53p8

This study aims to propose a polynomial distributed lag model to predict the Brazilian financial system default using macroeconomic variables. This model estimates coefficients that consider lagged effects of the explanatory variables in the response variable. The most commonly used models in stress testing (linear regression, time series and panel data) do not consider that the change in a variable has a distributed effect over the later periods. The model was estimated for the period from January 2004 to February 2016, in which a default was estimated using explaned macroeconomic variables related to interest, industrial production, unemployment and Ibovespa stock index. The proposed model presented a smaller error than the most commonly models used in the literature, indicating that its use is more adequate.