來華與出國觀光旅客需求預測模式建立之研究

Abstract

觀光事業向有"無煙囪工業"之稱,自政府於民國四十八年全力發展觀光事業之後,每年來華觀光旅客人數即不斷地成長,總人數更於民國78年突破200萬人次,這對國民外交的推動、政府外匯收入的增加有莫大的助益。在另一方面,自民國六十年政府開放國人出國觀光之後,由於國民所得的提高、台幣的升值及其它種種社經有利因素的影響,使得每年出國觀光人數穩定的成長,而在民國76年開放國人赴大陸探親之後,出國觀光人數更呈直線上升,這對於提高國家知名度以及展示國家整體經濟實力有極為明顯的助益。 來華與出國觀光旅客人數的多寡直接或間接影響本地觀光業者及政府相關單位對觀光業軟、硬體設施的投資以及整體策略的規劃,舉凡觀光旅館的興建、航空公司航線的增減、導遊人員的培訓以及政府駐外單位的配合措施,在在都有賴於對未來需求的精確預測,過於粗略或不當的預測,不僅將造成大量觀光資源的閒置與浪費,也將使得政府與觀光業者在這場日趨激烈的觀光事業競爭中處於極不利的地位。 本研究計畫蒐集並參考近十年來國內外學者在觀光旅遊預測模式方面的研究,針對來華與出國觀光旅客整體及各主要市場需求,尋找並建立適當之長短期預測模式。我們考慮下列六種模式:簡算法、單變量時間序列模式、轉移函數模式、時間趨勢模式(即迴歸模式)、指數平滑法以及計量經濟模式,同時利用各類模式選取準則如AIC、SBC等來選取最佳模式,或以平均絕對百分誤差(MAPE)以及根均方百分誤差(RMSPE)來評估各模式預測能力,並從中選出最佳模式。

Tourism industry is sometimes called the `non-chimney` industry. Our government has pushed forward the development of this industry since 1959. According to the official statistics, the number of tourist visits to the island grew rapidly in the period of 1959-1978 and steadily since then. On the other hand, the number of tourist visits from the island to other countries also grew steadily since 1971 when people in the island are allowed to do so, and rapidly since 1987 due to the increase of GNP, the rising exchange rate of NT dollars relative to the U.S. dollars, and other advantageous factors. Tourism demand directly affects the decisions of government and private sectors on their investments for the software/hardware facilities of tourism industry. If the demand forecast is seriously overstated, high levels of investments in transportation and accommodations can result. Conversely, an area which underestimates its tourism potential will develop less capacity, discouraging some tourists from visit there. Therefore, the ability to accurately forecast tourism demand can be very beneficial in their decision making. In this research project, we will investigate various statistical models such as regression models, time series models (including Box-Jenkins ARIMA models and transfer function models), double exponential smoothing and econometrics models to derive and construct forecasting models for long-term as well as short-term tourism forecasting of tour visits to/from the island. We also consider several well-known model selection criteria for the selection of these candidate forecasting models. These include AIC, BIC and others, from the model-fitting points of view, and MAPE (mean absolute percentage error), RMSPE (root mean square percentage error) from the prediction points of view. We hope that these research conclusions can be of great help to the policy decisions of government and private sectors for the future planning on tourism industry.