The importance of principal components in studying mineral prices using vector autoregressive models: Evidence from the Brazilian economy. (August 2019)

• Record Type: Journal Article
• Authors: de Souza Ramser, Claudia Aline; Souza, Adriano Mendonça; Souza, Francisca Mendonça; da Veiga, Claudimar Pereira; da Silva, Wesley Vieira
• Is Part Of: Resources policy. Volume 62(2019)
• Page Start: 9
• Page End: 21
• Publication Date: 2019-08
• Subjects: Multivariate statistics -- Brazilian economy -- Price analysis -- Mineral economics -- Mineral resources; Mines and mineral resources -- Periodicals; Ressources minérales -- Périodiques; Ressources naturelles -- Gestion -- Périodiques; Environnement -- Politique gouvernementale -- Périodiques; 333.8
• DOI: 10.1016/j.resourpol.2019.03.001
• Languages: English
• ISSNs: 0301-4207

Abstract

Abstract: This study examines the impact of the main Brazilian mineral commodity prices negotiated in trade balance using vector autoregressive models (VAR) in the Brazilian economy in a short-term period. VAR models were applied to the full original data and then to the data dimensionality reduced by principal components denoted by PC-VAR (principal component - vector autoregressive). In the study cases, Cholesky decomposition impulse response and variance decomposition were performed and compared in terms of short run co-movements to identify the most effective model. The applied PC-VAR methodology led to a significant reduction of variables, and similar co-movements were obtained in the short-term period when an impulse response was applied and compared to an unrestricted vector autoregressive. The proposed method also identified the most important variables that affect the other variables in the Brazilian economy and have the same co-movements. Highlights: Two methodologies were used to evaluate the behavior of the Brazilian mineral variables. The method proposed identified the variables and possess the same co-movements. The PC-VAR methodology applied presented significant reduction of variables PC-VAR is a practical tool And features a having a lower number of parameters The PC-VAR modeling showed a performance similar to the VAR modeling