Determining the Causal Effect of World Cocoa Price on Production of Cocoa in Ghana: A Modeling Approach

Models are created in this research to explain and forecast the effect of global cocoa prices on cocoa production in Ghana using a regression model with time series errors. The problem of price volatility is particularly decisive for Ghana due to her heavy dependence on cocoa exports for foreign exchange earnings. The experiment sought to determine whether the global cocoa price can aid in anticipating the future behavior of Ghana's cocoa production. Annual data from 1961 to 2010 were used to fit the model, with out-of-sample data from 2011 and 2012. The regression model with ARIMA(2,2,0) errors was deemed the "best" model for the production variable based on the results of numerous model adequacy techniques. The model was validated using the mean absolute percentage error (MAPE) as forecast accuracy metric. Therefore, the 'best' regression model's MAPE was 7.97%. The usual "best" ARIMA model, however, which was fitted to the production variable, predicted a MAPE of 16%. This demonstrates that when the production variable was modeled along with the world price using regression with ARIMA errors, the MAPE was reduced. Therefore, a regression model with ARIMA (2,2,0) errors is a more accurate statistical method than the ARIMA method for predicting Ghana's cocoa production.