Índice de condições financeiras, previsões condicionais e não causalidade

• Ano de defesa: 2019
• Autor(a) principal: Barros, Julio Cesar de Mello
• Outros Autores: [e-mail]
• Orientador(a): Lima, Elcyon Caiado Rocha
• Banca de defesa: Brandão, Antônio Salazar Pêssoa , Cunha, Alexandre Barros da, Oliveira, Luciano Vereda
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade do Estado do Rio de Janeiro
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciências Econômicas
• Departamento: Centro de Ciências Sociais::Faculdade de Ciências Econômicas
• País: Brasil
• Palavras-chave em Português: Economia; Política monetária; Modelos econométricos; Previsão econômica; Desenvolvimento econômico; Índice de condições financeiras; Previsão condicional; Modelo não-causal; VAR
• Palavras-chave em Inglês: Financial conditions index; Conditional forecasts; Non-causal model
• Área do conhecimento CNPq: CIENCIAS SOCIAIS APLICADAS::ECONOMIA::METODOS QUANTITATIVOS EM ECONOMIA
• Link de acesso: http://www.bdtd.uerj.br/handle/1/19342

The main goal of this study is to develop a methodology that allows us to obtain a Financial Conditions Index (FCI) that allows us to infer when the financial conditions of a given country are tight or loose, and what is the relationship between these financial conditions and the output gap. We intend to extend the work of Céspedes et al (2006), who use a causal VAR (autoregressive vector) model and the conditional forecast theory (Waggoner and Zha (1999) and Doan et al (1984)) to construct a Monetary Condition Index. (MCI). The described methodology will be extended to incorporate a large number of financial variables and also deal with the potential non-causality of VARs, which, as Alessi et al (2008) points out, is very common in economic models, especially when there are additions of financial variables. More particularly, we have developed a methodology that allows conditional forecasts to be made from purely non-causal models and use this methodology to develop a non-causal Conditional Financial Conditions Index (CFCI). In developing this methodology, we rely heavily on the methodology developed by Lanne and Saikkonen (2013) for forecasting using non-causal VAR models. The main results of our work point to the existence of non-causality in the model with financial variables, which aims to relate the movements of the output gap with the CFCI. The results indicate a high dynamic correlation between non-causal and causal ICFC and the product gap, even in more distant horizons (8 quarters). In addition, we find evidence that non-causal CFCIs are more effective at capturing turning points than their competitors, indicating some anticipation of business cycles. We also found that US financial conditions, as measured by our index, were tight during the 2007/2009 financial crisis period, corroborating what is presented by other authors.