Estudo de séries temporais de preços de petróleo, etanol e açúcar através da Transfer Entropy

Abstract

The Brazilian energy infrastructure in the transportation sector has unique characteristics, with a large participation of bioenergy. The Brazilian government obliges the addition of ethanol in gasoline and this percentage varies as a political mechanism of price stabilization, increasing or decreasing depending on the industrial capacity to produce ethanol, the price of oil. The price of sugar cane also has a role in this decision-making. The domestic market switches the destination of sugarcane to ethanol or sugar production depending on market prices. Much of the ethanol production is destined for the domestic market. While sugar, besides being a product of the Brazilian basic food basket, is a strong product of exports. Thus, sugar, ethanol and petroleum commodities have a very large price correlation in the Brazilian financial market. The correlation between the prices of ethanol, sugar and oil leads to the need to study this exchange of information and to evaluate the directionality of this flow of information. In the analysis of multivariate time series, a common subject of interest is the coupling between the variables. In order to study the transfer of information in the series of prices of these commodities we use the Transfer Entropy that allows, in addition to measuring the coupling force between the series, to identify the directionality of the coupling. As these commodities are highly influenced by the international market, we analyze the behavior of their time series of prices in the face of the global financial crisis that led to the collapse of Lehman Brothers Bank. In this way, we study these commodities for the periods before, during and after the Subprime Crisis. Our main objective in this study was to investigate time series of ethanol, oil and sugar prices using the Transfer Entropy method. The results show that the financial crisis affected the relationship between sugar, ethanol and oil commodity prices. Transfer Entropy was efficient in quantifying the flow of information between the time series of prices of these commodities and elucidating the directionality of the transfer. We highlight some inversions observed in the intensity of the flow of information, for example between sugar and oil, during the periods during and after the crisis, the behavior of Transfer Entropy reversed and the direction of oil sugar became bigger than the other direction. Between sugar and ethanol and between oil and ethanol there was also an inversion in the behavior of information transfer during the period during the crisis. Overall, the method detected the change in price behavior caused by the financial crisis. The advantage of the technique is, besides measuring the intensity between these couplings, to indicate the directionality of the interactions. Among the results obtained with the Multiscale Transfer Entropy method we highlight that between oil and ethanol, when analyzing the different time scales, we could not conclude which direction transfers more information. This feature may be related to policy interventions that lower the price of gasoline artificially, change the percentage of ethanol in gasoline, causing the relationship between ethanol and oil prices not to occur naturally.