Measuring core inflation in Brazil using a svar approach

• Autor(a) principal: Rabe, João Paulo De Faria Tavares
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/121889
• topic: Inflation; Core inflation; Svar; Brazil; Long-run phillips curve; Domínio/Área Científica::Ciências Sociais::Economia e Gestão

The main objective of this paper is to present a new measure of core inflation for the Brazilian economy. Different from the statistical and a theoretical measures usually used by the Central Bank of Brazil, the methodology is based on Quah and Vahey (1995)and is backed by the economic theory that the Phillips Curve is vertical in the long-run, there fore the core inflation is calculated as the component of the measure inflation that does not impact the output level in the long-run. This is an approach almost not explored in Brazil, and the results have shown that could be beneficial if included in the set of measures followed by the Central Bank. In order to calculate this core measure, two Structural Vector-auto regression models are used. Firstly, it is estimated the bivariate model proposed by Quah and Vahey (1995),in which the (log) difference of output level and the (log) difference of price level are use daiming to identify core and non-core shocks. The results, as discussed in the related literature, showed that both structural shocks have a similar pattern to what the theory identifies as positive demand and supply shocks. Further, in light of Bjørnland(2000) and Martel (2008), a commodity price index is added to better identify the shocks affecting the system. Both models point out that the measured inflation and the core inflation measures follow the same trend, whereas short-term inflation is mainly due to supply shocks. Although there seems not to be a consensus as to what is the best methodology to calculate a core inflation measure, hence the literature recommends that a core inflation measure should have some specific characteristics. Therefore, a comparison among the measures produced by the SVAR models and the ones typically used by the Central Bank of Brazil is conducted to evaluate these features. The results pointed out that among the core measure analyzed the only ones systematically unbiased are produced by the SVAR approach. Moreover, the SVAR methodology also showed more gains in tracking the inflation trend than the exclusion methods, whereas the core measure based on trimmed means–by construction -is the oewith the lowest error. Finally, the core measures with the best forecast(out-of-sample)performance are the SVAR trivariate system, the Ex3, and the P55.