Exploring the presence of nonlinear deterministic dynamics in commodity prices

Abstract Understanding the behavior of agricultural commodity prices is crucial for policymaking, forecasting, production, storage, investment, and risk management decisions. This study investigates whether agricultural commodity prices and volatility are generated entirely by linear stochastic processes or whether they contain evidence of low-dimensional nonlinear deterministic dynamics. We apply nonlinear time series analysis to empirically detect the underlying dynamics of commodity futures prices. Phase space reconstruction, nonlinear predictive skill, and permutation entropy measures are used to distinguish between linear stochastic and nonlinear deterministic dynamics. The results reveal evidence consistent with low-dimensional nonlinear deterministic dynamics in the structured component of commodity prices, which accounts for 82–98% of total price variation across commodities. These findings are suggestive of a substantial portion of observed volatility being generated endogenously by an inherently unstable market that may lack the ability to self-correct. The findings may have implications for commodity-price stabilization policies, although the form, timing, and effectiveness of any intervention require further investigation. They also point to the possibility of improving short-term price forecasts using nonlinear methods, although long-term forecasts remain challenging.