聯立方程式與單變量ARIMA模式之應用-以高雄港貨櫃量預測為例

The application of Simultaneous equations and univariate ARIMA model-forecasting of the container volume for Kaohsiung Port

蔡明君(Ming-jyun Cai)

指導教授 ： 謝浩明

國立中央大學/工學院/土木工程學系/碩士(2011年)

摘要

本研究主要目的為建構高雄港進出轉口貨櫃量的預測模式。實務發現進口貨櫃量為國內經濟活動的衍生性運輸需求，出口貨櫃量為國內與貿易國之經濟活動的衍生性需求。而國內經濟活動與貿易國經濟活動間彼此乃是息息相關，因此進口貨櫃量與出口貨櫃量間亦為彼此相互影響之聯立關係。轉口貨櫃量受航商運輸行為改變、大型化貨櫃船趨勢和子母船運輸方式盛行的影響。基於大型化貨櫃船的運量需求，貨櫃母船靠泊於貨櫃量運輸需求大的港埠，鄰近地區的貨櫃量運輸需求則先於集貨港進行集貨動作，再利用子船前往裝載回母船靠泊港，以完成整體運輸行為。隨著大陸地區內需力量的激增，龐大的貨櫃量需求吸引下，航商逐漸將運輸路線轉移至大陸地區深水港埠，我國進出口貨櫃量之需求雖不及大陸地區，但對航商而言仍是可觀的貨櫃量需求，加上高雄港擁有較佳的港埠裝卸效率、進出港船舶艘次的頻率頻繁及港埠相關營運政策有利於航商等因素吸引下，均有利於高雄港爭取區域內集貨港的地位，故轉口貨櫃量受進出口貨櫃量之影響，形成進出轉口貨櫃量三者的聯立關係。港埠運量預測ㄧ直以來皆為港埠投資計畫中重要的環節，準確的港埠運量預測能有效的幫助營運當局掌握未來貨櫃量之需求，進而擬訂最適的投資計畫，減少不必要的國家建設開支浪費。本研究利用進出轉口貨櫃量三者的聯立關係，以歷史資料建構出進出轉口貨櫃量彼此聯立影響行為之模型，再利用ARIMA模型產生外生變數預測值，最後將外生變數預測值代回聯立模型內，以求得進出轉口貨櫃量預測值。研究結果發現，影響進口貨櫃量之顯著變數為出口貨櫃量，影響出口貨櫃量之顯著變數為進口貨櫃量和香港工業生產指數。影響轉口貨櫃量之顯著變數為出口貨櫃量、境外轉運中心貨櫃量、平均在港停泊時間及平均進出港船舶艘次。最後加入海西經濟特區及台北港加入營運後之實務討論，修正研究所得之高雄港貨櫃量之預測值。

關鍵字

出口貨櫃量 ； 預測 ； 轉口貨櫃量 ； 聯立方程式 ； ARIMA模型 ； 進口貨櫃量

並列摘要

The main purpose of this research is to build a forecasting model for the container throughput (import/export/transshipment) of Kaohsiung Port. Practice shows that import container volume is a derivative transportation demand which is related to domestic economic activity while export container volume reflects the derivative demand of economic activity between Taiwan and other trading countries. Domestic economy is closely connected to other trading countries, so import container volume and export container volume influence each other in a mutual and simultaneous way. The volume of Transshipment container is effected by the transportation behavior change of container carriers, large-scale trend for container ships and flourishing of lighter aboard ships; in response to the freight volume demand of large-scale container ships, that is, mother vessels berth alongside ports with bigger container traffic demands. For the container traffic demands in neighboring areas, first consolidating in feeder ports, then utilizing barges to transport the goods to the ports where mother vessels are berthing, hence, to complete the entire transportation. As the surge of China’s domestic demand, attracted by the tremendous demand for containers, the shipping companies gradually relocate the ship routes to deep-water ports in China; Taiwan’s import/export container demand is not as big as that of mainland China, but for shipping companies, it is still a considerable demand. Furthermore, considering the factors that Kaohsiung Port is more efficient in loading and unloading, has frequent incoming and outgoing ships and port related policies favor shipping companies, all of these are beneficial for Kaohsiung to win the position as top regional feeder port. Thus, the transshipment container volume is affected by import and export container volumes and forms simultaneous relationship among import, export and transshipment container volume. The forecasting of port traffic has always been an important factor in port investment plan and exact port traffic forecasting can effectively help port operators to know future container demands, hence, designing the most suitable investment plan can reduce unnecessary national construction expenses. This research utilizes the simultaneous relationship among import, export and transshipment container volumes and builds a simultaneous influence behavior model of import, export and transshipment with historical data. Using ARIMA model to gain estimated exogenous variables and substituting estimated exogenous variables in simultaneous equation to gain the forecasted value of container throughputs. The research shows that the significant variable for import container volume is the export container volume; the significant variables for export container volume are import container quantity and index of industrial production in Hong Kong. The significant variables for transshipment container volumes are export container traffic, container volume in offshore transshipment center, average berthing time and the average number of incoming and outgoing ships. Finally, adding Special Economic Zone Taiwan Strait West bank as well as Taipei port operation into the empirical discussion, and modifying the estimated value of container traffic in Kaohsiung Port calculated in this research.

並列關鍵字

forecasting ； simultaneous equations models ； transited containers ； export containers ； import containers ； ARIMA models

參考文獻

25. 郭英峰和陳邦誠，「以單變量ARIMA模式、類神經網路、灰色GM(1,1)模型預測高雄港貨櫃吞吐量」，台大管理論叢，第17卷第2期，107-132頁，民國96年。

29. Badcock, M. W., and Lu, X., “Forecasting inland waterway grain traffic,” Transportation Research Part E, Vol.38, pp. 65-74, 2002.

31. Chavez, S. G., Bernat, J. X., and Coalla, H. L., “Forecasting of energy production and consumption in Austurias (northern Spain) ,” Energy, Vol. 24, pp. 183-198, 1999.

32. Cho, V., “Tourism forecasting and its relationship with leading economic indicators,” Journal of Hospitality and Tourism Research, Vol. 25, No. 4, pp. 399-420, 2001.

33. Clark, X., Dollar, D., and Micco, A., “Port efficency, maritime transport costs, and bilateral trade,” Development Economics, Vol. 75, No. 75, pp. 471-450, 2004.