Navigating the Flow: Unveiling Directional Information Transfer in Commodity Markets With Transfer Entropy and Moving Window Analysis

Abstract

This study examines directional information flow in commodity futures markets using transfer entropy (TE) and Granger causality (GC) over 21.5 years. Analyzing 12 major commodities through rolling windows of 20, 60, 120, and 240 days, we compare linear versus nonlinear transmission mechanisms across different market conditions. The results show that while GC captures persistent linear relationships with monotonically increasing detection rates, TE reveals complementary nonlinear dependencies following nonlinear patterns. During crisis periods, network density increases significantly, with the COVID-19 pandemic producing pronounced nonlinear effects where TE substantially exceeds GC. Energy commodities dominate linear channels, while agricultural commodities emerge as central nodes in nonlinear networks. Period-specific analysis reveals regime-dependent transmission: postcrisis periods return to linear relationships, while commodity-specific shocks activate different nonlinear pathways than systemic crises. The findings demonstrate that comprehensive commodity market analysis requires both methodologies to capture the full spectrum of information transmission mechanisms.