以潛在類別模式模化計程收費下高速公路駕駛人之運具及路線選擇移轉行為

Abstract

為落實使用者付費之公平精神，交通部台灣區國道高速公路局擬將現行的計次收費制改為計程收費制，然現行高速公路使用者多數屬短程免付費旅次(佔高速公路總旅次之65%)，為避免大量短程旅次受收費衝擊改變路線造成平面交通擁塞，國道高速公路局擬加入免費里程機制。為了解高速公路駕駛人因應未來政策的選擇行為，以及免費里程機制與通行費率的政策效果，本研究設計敘述性偏好實驗(stated preference experiment)訪問臺北、臺中及高雄三大都會區高速公路駕駛以得知不同收費情境下之運具及路線轉移。又為了解高速公路駕駛人的個體偏好差異，本研究應用潛在類別模式(latent class model, LCM)模化高速公路使用者的選擇行為。

潛在類別模式的推估結果顯示旅次長度、居住地區及年齡為影響分群之關鍵變數，並將高速公路使用者分為三群：(1)一般高速公路使用者(2)短程─高所得及(3)長程─低所得，不同分群對旅行時間及旅行成本的偏好有顯著的差異。此外，性別、所得及教育程度在上述分群當中亦有顯著差異。本研究依據模式推估結果設計不同通行費率及免費里程之情境作敏感度分析，分析結果發現，免費里程機制可以降低短程旅次受通行費率衝擊的程度，而長程旅次在各類免費里程情境下受通行費率影響的敏感度最大。最後，長程─低所得族群的社經背景多屬較偏遠地區居民，本研究建議提升偏遠地區公共運輸服務水準以滿足該群體受衝擊後轉移至公共運具之前在需求。

For the principle of pay-as-you-go, Taiwan government is planning to switch the current entry-based toll system to a distance-based toll system. However, to avoid the most of short-range and free-of-charge trips, accounting for approximately 65% of total freeway trips, diverting to use of local roadway systems and causing serious congestion on the systems, several distance-based schemes with and without daily credit of free-mileage are then proposed. Without knowing the mode and route choice behaviors of freeway drivers, it is impossible to fairly assess the toll schemes. Based on this, this study aims to develop the mode and route choice model of freeway drivers so as to investigate the potential impacts to them under various distance-toll schemes. To acknowledge the heterogeneity in the preferences of freeway drivers, latent class models (LCM) are developed based on the stated-preference designed questionnaire surveys on Taipei, Taichung and Kaohsiung metropolitans.

The estimated results of the LCM show that trip length, residential area, and driver age are three key segmentation variables, which classify freeway drivers into three groups: 1) Ordinary freeway drivers; 2) High income with short trip drivers; and 3) Low-income with long-trip drivers. It is found that not only the preferences of travel cost and travel time but also gender proportion, average income and average education levels significantly vary across groups. Sensitivity analyses on various toll rates and various daily credits of free mileage are conducted and compared. The results show that the mechanism of daily credit of free mileage can effectively curtail the impact of drivers with short trips; however, drivers with long trips are more sensitive to toll rates. Additionally, since most of drivers in the group of low-income with long trips live in rural areas, suggesting that to improve the public transportation service in those areas is imperial so as to reduce the negative impact caused by the distance-toll scheme.