Avaliação do risco antes e após reeleição de Dilma Roussef: uma análise por meio de redes de correlação

Abstract

Brazil went through a political and economic crisis period from 2013 to 2016. The electoral results, in 2014 interfered in the local stock market risk. Data from BM&FBOVESPA suggested an increase in the local stock market volatility. Similarly, loss in the most of the companies` market value. Recession and crisis marked the following years. However, loses in financial and consumer`s sectors seemed to be lower. In this way, we asked if also the systemic risk would have grown, if the central companies and sectors would have changed and what would be the minimum variance and systemic risk portfolios. A dynamical Minimum Spanning Trees (MST) analyzed the systemic risk. I built 48 mothly MST´s from October 2012 utill September 2016 and analyzed the first four moments of the distance correlation matrix. I proposed two MST`s to verify the optimal portfolios and changes in the central companies and sectors: one for the two years before the re-election of Dilma Roussef and another for the two years after. I compared the central vertices and the number of clusters by hierarchical clustering. By observing the MST length`s shrinking and the increases in the correlation level, I discovered that the systemic risk grew through the following six months after the electoral results. Changes in the inclination of the distance correlation`s asymmetry and kurtosis suggested expectation and uncertainty changes. The hierarchical clustering revealed that the central stocks before the reelection were BBDC4- Bradesco Bank and EVEN3-EVEN Construction Company. After that, BBCD4 and ITSA4 – Itau Bank remained as the most central stocks. Construction sector left to be an important sector and stood by ITSA4 periphery. I suggested 8 clusters for each period as the available number for portfolio diversification for systemic risk reduction. The 20 lowest closeness centralities indicated the minimum variance portfolios for both periods.