Price Dynamics in Commodity Market : A Comparison between European and US Markets

This paper tests for the presence of long memory and nonlinearity in returns and volatility for six agricultural futures daily prices series, three traded on MATIF Euronext (Wheat, Corn & Rapeseed) and three traded on CBOT (Red Winter Wheat, Corn & Soybean) over the period from 2000 to 2010. If the price dynamics on the CBOT market seems to be described by classical ARMA-Garch modeling, time series dependences on the MATIF market do not appear to be fully described only by short-term dependences. Using various criteria such as Hurst exponent, correlation dimension and BDS test, the result suggests the presence of long memory for the European market. However, it appears that low fractional order of ARFIMA-type or FiGARCH-type models can explain, but not all, the observed nonlinearity. Nonlinearity could be influenced by regime shift. Subsequently we screened series of structural breaks influence on volatility. Breaks seem be caused by temporary public intervention on the market. Although we cannot fully accept the assumption of independence for all filtered series, serial dependences on the MATIF series appear to be largely explained by structural changes on volatility related to the policy of public intervention in the market