Empleo del comportamiento estacional para mejorar el pronóstico de un commodity: el caso del mercado internacional del azúcar

Abstract

This paper studies the behavior of seasonal patterns in the international sugar price. Using seasonal unit root test and a monthly sample from 1989 to 2010, a non-stationary stochastic seasonal pattern was observed. This pattern implies that a “summer” could become a “winter”, a result that had not been previously documented for this market. On the other hand, using these findings, our results show that is possible to create an AutoRegressive Moving Average (ARMA) model that out-performs other approaches that do not take in account this kind of seasonality when forecasting the sugar price.