Modelagem da PD forward looking: uma análise de impacto dos fatores macroeconômicos na inadimplência bancária nacional

• Autor(a) principal: Simões, Robson de Souza
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional do FGV (FGV Repositório Digital)
• Texto Completo: https://hdl.handle.net/10438/32979
• topic: Expected credit loss; Default rate; Macroeconomic factors; Multivariate regression model; Perda esperada de crédito; IFRS 9; Taxa de inadimplência; Fatores macroeconômicos; Modelo de regressão multivariada; ARIMAX; Economia; Créditos - Avaliação de riscos; Bancos - Brasil; Contabilidade - Normas; Inadimplência (Finanças); Análise de regressão

The new expected credit loss model defined by the international standard IFRS 9 aims for banks to refine their provision calculation estimates considering both the losses already incurred and those expected in the future. In one of the passages of the IFRS9 it is mentioned that studies must be conducted about the influence of macroeconomic variables on the models used in banks' provision calculations. By identifying those macroeconomic factors that have statistical significance for the historical changes in a portfolio, it is expected that banks will be able to integrate them and thus improve their probability of default calculations. A study that evaluates the relationship between macroeconomic variables and Brazilian default from an IFRS9 perspective will provide inputs that will help financial institutions gain a better understanding of this relationship to adopt the regulatory requirements. The aim of this paper is to demonstrate some methods that assess this relationship and how to proceed to incorporate macroeconomic fluctuations in the default rate. The macroeconomic factors studied are GDP, the IPCA, the exchange, the unemployment rate and the Selic rate. The default rate is based on data from the national credit portfolio of individuals (PF) provided by BACEN, where the default criterion occurs when there is a delay greater than or equal to 90 days in the payment of credit obligations, and the period used was from Jan/2019 to Jan/2022 Multivariate regression model and ARIMAX model are used to explain the relationship between macroeconomic factors and default rate. The results show that GDP is the main statistically significant macroeconomic factor for detecting changes in default rate and consequently in the calculation of expected credit loss