摘要：Cocoa beans are produced in equatorial and sub-equatorial regions of West Africa, Southeast Asia and South America. These are also the regions most affected by El Nino Southern Oscillation (ENSO) -- a climatic anomaly affecting temperature and precipitation in many parts of the world. Thus, ENSO, has a potential of affecting cocoa production and, subsequently, prices on the world market. This study investigates the benefits of using a measure of ENSO variable in world cocoa price forecasting through the application of a smooth transition autoregression (STAR) modeling framework to monthly data to examine potentially nonlinear dynamics of ENSO and cocoa prices. The results indicate that the nonlinear models appear to outperform linear models in terms of out-of-sample forecasting accuracy. Furthermore, the results of this study indicate evidence of Granger causality between ENSO and cocoa prices.
