Forecasting Fish CPI-Using Box-Jenkins Time Series Model

摘要

The paper is an application of the Box-Jenkins (B-J) univariate modelling approach to make short run forecasts of the Consumer Price Index for fish (Fish-CPI). After giving a general background of the fishing industry and the reason for using such techniques to forecast the Fish-CPI, an overview of the univariate B-J procedure is given. Two models were constructed. After evaluating these models, one was selected to generate a 6-month forecast. The 6-month forecasting error is analyzed statistically. An extended 12-month forecast was then made using the forecast correction function from the forecast error analysis. The result indicates that the model, given its simplicity, yields good results over a large part of the forecast horizon.