Analysis of the nonlinear relationship between commodity prices in the last two decades

Abstract

We propose the construction of a network to study the correlation among price indices of different commodities, by using the Multifractal Cross-Correlation method proposed by Podobnik and Stanley. This estimator, based on the method Multifractal Detrended Fluctuation Analysis, is effective for self-similar signals with characteristics such as those we here analyze. We construct different networks for time periods between 1991 and 2012. Each node represents a commodity group and the links are the cross-correlation between nodes. We study the evolution of these networks from January 1991 to April 2012. The results show that after 2007, high connectivity arises between the nodes of the matrix. We conjecture that this is a consequence of the cash flow from equities and real estate markets to the commodity market due to the subprime mortage crisis.