Time varying correlation: key indicator in finance

An important component of asset allocation decisions or exotic derivatives pricing is represented by the estimation of the correlation structures observed in various markets. In commodity markets where prices result non stationary and returns are only mean stationary, a time varying measure of correlation has to be used. According to the prevailing literature correlations among different markets are higher during recessions than during growth periods. Portfolio managers who aim to shield investors from stock markets declines used to invest in commodities which historically had been found to be poorly correlated with stock markets. In the last five years correlations between commodities returns have dramatically changed. According to some analysts investors seem to ignore cracks in the global economy which may cause structural changes in the relationship between different markets. Some investors view the recent disparity in the performance of commodities and equities markets as an indirect endorsement of the government’s efforts to shore the domestic economy without stoking inflation. The present work aims to study the correlation structures of returns in various markets using different dataset and methodologies. Starting from a GARCH (p,q) variance analysis for several commodity markets we estimate the Dynamic Conditional Correlations over the period 2000- 2014 for a set of commodities and three different indexes (stock index, commodity index and bond index). We found unexpected low or no meaningful correlations between energy commodities returns, natural gas prices result no correlated at all to crude oil prices; in particular natural gas prices result to be poorly correlated to the other considered commodities, and to the various equity and bond indexes. Some very low correlations are also found between the gold market and crude oil and bond markets. Some interesting structural breaks were also found in the correlation dynamics. JEL Classification Numbers: C32 , Q40.