Palm oil price model of Indonesia market

Abstract

Time series models, i.e. AR, ARMA or ARIMA model, are widely used to model the commodities prices. Such models perform well in term of the price forecast. Another approach, modeling using stochastic differential equation is essential to value derivative, e.g. option on commodities. We study modeling palm oil spot prices traded in Indonesia market by the potential diffusion model, which is close in spirit to a mean-reversion diffusion model, with a more general drift term. Such a model allows to have multiple attraction regions and allows the mean-reversion rate to be a continuous function of the distance to the mean price level. On the basis simulation study, we compare performance of this model to classical models using stochastic differential equation: Geometric Brownian Motion and mean-reversion diffusion. We apply some statistical tests and forecast the future spot prices to investigate performances of those models. Our study shows that the potential diffusion model performs better than the geometric Brownian motion and the mean-reversion diffusion models.