Aplicação da teoria de cópulas bivariadas na modelagem de dependência do retorno de ações da BM&FBOVESPA

Abstract

The financial time series presents peculiar characteristics that make the modeling of the structure of dependence of these data, important for the understanding of questions in finance. In this context, the copulas theory is an important tool in the multivariate analysis of financial series, thanks to its flexibility to construct multivariate distribution functions that reproduce nonlinear dependencies. Thus, this work has the objective of applying the conditional copulas methodology with fixed parameters to model and evaluate the relationship between the most representative actions of the main sectors acting in the Stock Exchange and Commodities of São Paulo in the period from January 3, 2011 until June 21, 2017. The results point to the existence of a considerable dependency ratio between the shares of Bradesco, Itaú Unibanco, Petrobras, Vale and Renner, in general, they relate a little more pronounced in periods of high prices than in periods of low prices. There is also a low degree of association in the shares of Ambev and BR Foods with the others, since the consumption of basic goods presents stability, even in times of economic recession and thanks to the planning of these companies. It should be noted that the particularities present in the sectors of each stock and the effects of the unstable domestic environment, verified during this period, can explain substantially the pattern and magnitude of dependence measured between these actions.