Dynamics of Commodity Prices. A Potential Function Approach with Numerical Implementation

摘要: In the present analysis a nonlinear model is discussed in order to capture presence of several forces acting commodity markets and difficulty disentangle their relative price impacts. Global have experienced significant swings recent years. Analysts offer two explanations: market speculative expectations, not mutually exclusive. Commodity prices seem indicate that various factors are very complex way. We start from one specific feature: clustering phenomenon, which tendency concentrate number attraction regions, preferring some values over others. Commodities process becoming mainstream. The mean-reverting class diffusion models able phenomenon multiple regions. potential function approach modelled as governed by function. A fundamental step fit multimodal density invariant distribution. postulate parametric form distribution framework finite mixture Expectation-Maximization algorithm. procedure for identifying estimating parameter provided. Applications crude oil soybean essential characteristics data remarkably well. An underlying assumption long-term volatility do change with time. New conditions new regions can form, changing shape magnitude volatility. investigate changes potential, reflects equilibrium levels (attraction regions) hence conditions. allows generate copies observed series same distribution, useful applications requiring large independent trajectories. goodness-of-fit test SDE numerical implementation