台灣地區海洋貨櫃運輸商品運量預測方法之研究

Abstract

本研究利用財政部關稅局的統計資料，建立預測海洋貨櫃商品運量之模式，同時將結果運用於實務上。以往關於海洋貨櫃運量預測之研究，其進出口運量之資料來源大都來自交通部各港務局的統計。財政部關稅局之進出口運量與價值統計月資料，係根據中華民國商品標準分類，各商品資料共有二位碼、四位碼、六位碼、八位碼及十位碼等五種分類，較交通部的統計更為翔實可靠。本研究使用其二位碼共97個商品章別1989.01至2001.03之進出口月資料，以平均貨櫃運量多寡篩選適合海洋貨櫃運輸的商品章別，並根據篩選結果，把各商品章別之進（出）口價值，用線性迴歸、指數模式以及42種時間數列方法中配適度最佳的模式，轉換為進（出）口運量預測，並比較各方法的優劣。結果顯示進（出）口價值與進（出）口運量存在高度正相關，可以互為替代變數，同時按商品運量及貨櫃運量計，大部分的商品章別經轉換的運量預測結果優於不經轉換之預測。依上述方法建立長期預測模式，進行台灣地區未來20年之商品與貨櫃運量預測，並與台北港興建暨營運等各大計畫之預測比較。結果顯示本研究所採用的預測方法具有實用性，可用於港埠興建暨營運規劃之參考。

Using the statistics compiled by Directorate General of Customs, Ministry of Finance, container water transport forecasting models were built, and the forecasting results could be applicable for practical purposes. Formerly, the statistics compiled by harbour bureaus, Ministry of Transportation and Communications, were often adopted in many researches. According to The Standard Classification of Commodities of the Republic of China, there are five sorts of codes to classify the monthly data of the import and export value and quantity. The MOF data seems to be more reliable and detailed than that of the MOTC. The thesis utilized the MOF data from 1989.01 to 2001.03 in two-digit code and 97 chapters of the commodities. The import(export) values were transferred to the import(export) quantities by using various transformation models, such as regression model, exponential model, and 42 kinds of time series models. The best goodness-of-fit models are selected. The results revealed that the import(export) value and the import(export) quantity were positively correlated, and can be proxy variable for each other. In the container traffic or the commodity transport aspect, the most part forecasted by transformation models are more accurate than those from the direct forecasts. The long-term forecasting models were built in the same way, and compared with forecasting results of several large-scale plans. The results of this thesis would be applicable for harbour constructions and operations.