差別費率策略下高速公路假日旅次之旅運選擇行為

Abstract

在臺灣由於高速公路於連續假日期間的交通壅塞問題日益嚴重，興建道路擴充容量之方式除了成本高昂外，也難以追上需求量增加之速度。而在現今高速公路收費系統以全面轉為計程式的電子收費系統（Electronic Toll Collection，ETC）後，使用差別定價此類交通管理措施的門檻已大幅下降。因此為減輕高速公路連續假日期間之交通壅塞，本研究旨在探討尖離峰差別定價對用路人選擇行為之影響，期望能藉此提出適當之差別定價方案，平衡連續假日尖離峰流量差距。 本研究以顯示性及敘述性偏好法進行問卷情境設計，採用了多項羅吉特模式及巢式羅吉特模式建構運具及路徑選擇行為、連續羅吉特模式建構出發時間選擇行為。調查採用互動式電子式問卷，並且於各主要連續假日後，在高速公路服務區進行調查，以鎖定目標對象。由模式推估結果知，巢式羅吉特模式雖將使用高速公路但出發時間不同之方案歸納為同一巢，但並無法得到較多項羅吉特模式更佳之推估結果，顯示不同出發時間之替選方案間之關聯性並不高。另外，從模式推估結果知，假期愈長之旅次，用路人愈不願改變其運具、路線及出發時間；但具有使用不同運具或路線經驗者，則願意改變之偏好較高。至於連續羅吉特模式之配適值偏低，無法完整模擬出整個連續假日期間之每日尖離峰分佈，僅能顯示大致趨勢。 而根據問卷情境資料分析的結果來看，尖峰時段定價從當前費率改為5倍費率之後，尖峰流量並沒有產生明顯變化，顯示定價策略對出發時間的影響小，因此建議改採用其他交通管理措施代替。

In Taiwan, freeway systems are usually seriously congested during consecutive holidays. The network expansion plan is hard to catch up with the increase of travel demand. However, since the freeway tolling system has been electronized, differential pricing strategies to effectively spread out peak traffic demand becomes viable nowadays. In order to relieve freeway traffic congestion during consecutive holidays, this study aims to examine the effect of differential pricing strategies to driver behaviors. Based on this, this study uses of revealed and stated preference questionnaire survey and respectively employs multinominal Logit model (MNL) and nested Logit model (NL) to modeling mode and route choices and continuous Logit model (CL) to modeling departure time decisions. An electronic interactive questionnaire is designed to face-to-face interview drivers who have recent experiences in using freeway systems during consecutive holidays at freeway rest areas. The estimation results show that the nested Logit model (composing of freeway usage with different departure time alternatives into the same nest) fail to perform better than multinominal Logit model, suggesting the low correlations among different departure time alternatives. Additionally, the estimated MNL model suggests that the longer holiday is, the lower intention to change travel behaviors is. However, those who have experiences in using alternative routes and modes are of higher intentions to change. Moreover, due to the low goodness of fit of the estimated continuous Logit model, the explanatory power is limited and the detailed departure time probability density curve is hard to be portrayed, but only the trend of departure time preferences. According to the scenario analysis under current toll rate and five times toll rates, the peak traffics are not remarkably different, suggesting the effect of differential pricing strategy on the departure time decisions is low and other countermeasures should be attempted instead.