Nonlinear dependencies on Brazilian equity network from mutual information minimum spanning trees

TL;DR

This study uses mutual information minimum spanning trees to analyze nonlinear dependencies in the Brazilian equity market, revealing increased risk and nonlinear interactions during political transition periods.

Contribution

It introduces the application of mutual information minimum spanning trees to capture nonlinear dependencies in financial networks, highlighting their robustness over linear correlation methods.

Findings

Mutual information networks show higher robustness and risk indicators.

Nonlinear dependencies increase during political transition periods.

Network structure correlates with stock performance.

Abstract

Mutual information minimum spanning trees are used to explore nonlinear dependencies on Brazilian equity network in the periods from June/01/2015 to January/26/2016, in which Brazil was under the government of President Dilma Rousseff, and from January/27/2016 to September/08/2016 which includes the government transition from President Dilma Rousseff to President Michel Temer. Minimum spanning trees from mutual information and linear correlation between stocks returns were obtained and compared. Mutual information minimum spanning trees present higher degree of robustness and evidence of power law tail in the weighted degree distribution, indicating more risk in terms of volatility transmission than it is expected by the analysis based on linear correlation. In particular, a remarkable increase of stock returns nonlinear dependencies indicates that the period including the government transition is more risky in terms of volatility transmission network structure. Also, we found evidence of network structure and stock performance relationship. Besides, those results emphasize the usefulness of mutual information network analysis for identification of Financial Markets features due to nonlinear dependencies.