GARCH-Type Models on the Volatility of Indonesian Cocoa’s Spot Price Returns

Since Indonesia is one of the largest cocoa producers in the world, cocoa has important role in Indonesian foreign exchange revenues through export. Meanwhile, as part of plantation commodity, cocoa’s price also has volatility nature. The ability of GARCH-type models in measure price volatility has been famous. These models have been applied properly for analyzing prices which demonstrate time varying level of variance. This study has two aims, to examine the predictability of GARCH-type models (ARCH, GARCH, GARCH-M, EGARCH, and TGARCH) on the cocoa’s returns volatility and to determine the best predictability model among the significant GARCH-type models. Two independent variables used in this study are the residual from the mean equation and volatility of error variances in the previous periods. The prices used are spot price series in periods of January 2005 to June 2011 from BAPPEBTI (Indonesian Commodity Futures Trading Regulatory Agency – CoFTRA). The results show that GARCH-M and EGARCH models are the best prediction models based on the error statistic criterions, RMSE, MAPE and MAE