ABSTRACT
This study provides abalone producers and consumers with price forecasting models exhibiting better performance for five classes of abalone prices based on shell size to help farmers predict abalone shipments and enhance the economic activity of consumers. First, the autoregressive-moving average (ARMA) models of abalone producer prices based on information criteria are selected. Second, the best ARMA model using out-of-sample data based on the mean squared error and mean absolute error is determined. Finally, this study compares the predictive accuracy of the better ARMA model and other ARMA models, using a modified Diebold–Mariano test. Higher forecasting accuracy is exhibited by the AR (1,3) model for 8 and 10 abalones per kilogram, the ARMA (1,0) and ARMA [(3),(1)] models for 13 abalones per kilogram, the ARMA (1,0) and ARMA [(1),(2)] models for 15 abalones per kilogram, and the ARMA (1,0) and ARMA (0,1) models for 20 abalones per kilogram.
Acknowledgments
The authors wish to thank Bongu Matthieu Mongolu and Odei Isaac Kwapong for their invaluable help with revisions.