Aplicação da metodologia VAR no mercado de ações internacionais

Abstract

The constant global economic crises that occurred during the 20th century created cycles of prosperity, and countries tried to control them to minimize recessionary outcomes. Therefore, the objective of this research was to analyze the stock indices of the most developed and industrialized countries of the world that are part of Group 8. Thus, we sought to analyze how stock indices of the United States, Japan, Germany, Canada, France, Italy, the United Kingdom and Russia behave in the short term. We estimated a VAR (1) autoregressive vector model, that is, a lag, identified by the Akaike Information Criterion (AIC = 30.04844) and Bayesian Information Criterion (BIC = 30.26366). The indexes were ranked according to their degree of exogeneity, classified as follows: index of France (CAD 40), Japan (NIKKEI), United States (NASDAQ) and United Kingdom (FTSE_100). Granger's causality resulted in the indices having a causal and bicausal relationship, which means the dependence between the variables in the study. From this result, the impulse and response function is used only in the variables that have the bicausal relationship, which are: France, represented by the CAD 40 index, and the NASDAQ, from the United States. In this way the test identified as a shock in the CAD 40 and influenced the NASDAQ behavior when there was an abrupt fall in the value of the index, but its behavior stabilizes in the short term. The decomposition of variance resulted in the percentage that each index has in the formation of the other, its degree of self-explanation over time. Therefore, it was concluded that, because they are indices of developed countries, they do not depend on the economies of the countries under study, with the exception of France and the United States.