Forecasting the Import Price of Hardwood Bleached Kraft Pulp Using a Time-Series Model

Abstract

Managing the import price risk of hardwood bleached kraft pulp (Hw-BKP) is a major concern of the Korea paper industry due to high price volatility. The primary objective of this study is to develop forecasting models for the import price of Hw-BKP as a tool to reduce price risk. Six different types of extended autoregressive models with various rolling window size were evaluated based on multiple accuracy measures. Results show that forecasts from the models with exogenous and monthly dummy variables are more accurate than the others. An autoregressive model with exogenous and monthly dummy variables model (ARMX) using a window size of 36 months outperforms the other models. When applying the model, it is possible to reduce the mean absolute error (MAE) by 24.6% and the root mean square error (RMSE) by 35.3% compared to a naive forecast model.