定期海運貨櫃航線運量影響因素分析與預測

Abstract

運量需求是航商在開發、調整航線以及船舶調度決策的重要依據，加上近期定期海運貨櫃運量需求變化劇烈，如何能事先預測需求變化以面對市場的快速變動並即時反應成為航商營運的關鍵課題之一。在航運業界不乏豐富經驗的經營管理者，能夠事先判斷到需求量之變動趨勢並適時的因應。然而，經驗需時間累積，且不易說明傳承，本研究擬以學術之定量預測方法，採用統計理論，試圖找出影響運量變化的主要因素以及預測貨櫃需求量之方法，事先預估短期航線運量之變動，以利航商即時因應。

基於上述研究背景與動機，本研究之目的主要在探討影響遠東出口至北美地區之定期海運貨櫃航線運量之重要因素以及建立預測航線運量之模式。研究方法主要以迴歸方法探索影響定期海運貨櫃航線運量之因素並以時間序列之ARIMA模式建立航線運量預測模式。研究結果顯示，在航線相關各國之經濟指標之中，經GDP平減指數調整之美國國內生產毛額、香港躉售物價指數以及日本工業生產指數對於航線運量有顯著之影響。而建構ARIMA模式過程發現，由於遠東出口至北美區之定期海運貨櫃航線運量具有季節變動特性，經測試結果，使用經季節調整後之航線運量時間序列所建構之ARIMA模式較原始航線運量之時間序列模式之預測誤差為小。最後本研究以上下限10%之誤差為區間提供短期(2011年四季)之預測航線運量以提供相關航商參考。

Cargo demand is an important basis for implementing or adjusting relevant shipping service. Recently, the demands of Liner shipping containers changed drastically. Therefore, how to cope with the rapid changing market and take proper actions in response to those changes immediately becomes one of the key issues for carriers. Experienced managers can predict changes in cargo demands in advance, and respond to those changes in a timely manner. However, establishing that kind of experience takes time and is hard to clearly articulate the rules to their successors. Thus, the current study adopts quantitative models to predict cargo demands for a short-term period, which is expected to provide Carriers as a reference to adjust their operating policies whenever necessary. The purpose of this study focuses on examining factors that influence the container volume of a liner shipping route and developing a model to reasonably forecast changes of container route volume. Regression analysis methods were applied to find out factors that influence the liner shipping container volume exported from Far East region to North America region and an ARIMA model was developed to forecast future volume changes. Results indicate that among many selected Macroeconomics indicators, adjusted GDP of USA、Wholesale Price Index of Hong Kong and Industrial Product Index of Japan are the significant influential factors. Moreover, the forecast errors of the generated ARIMA model are relatively smaller if seasonal adjusted container volume time series data is used. Lastly, this study provided four seasons of year 2011’s container volume forecast with plus/minus 10% error range for Carriers’ reference to adjust their policies.