Abstract: The article studies the correlation structures of a large panel of agricultural commodities prices between January 1990 and February 2014. We use a various collection of mathematical and statistical methodologies (estimated correlation matrix and principal component analysis) to capture these correlations.
Our results show that there exist different degrees of correlation between commodities. We also demonstrate, through data mining analysis, that there are hidden correlations between some commodities. Indeed, some commodities’ price behaviours are very similar in trend. Our results contribute to a better understanding of agricultural prices’ behaviours by producers, investors and market intermediaries. The results contribute to a more efficient strategic asset allocation process within agricultural markets.18