10. Modelling Users’ Willingness to Pay for Different Attributes of an Improved Transportation System
Date of Start

May 2005
Indian Institute of Technology, Kharagpur (R)


Scope and Objectives

The broad objective of the work is estimation of users’ willingness-to-pay (WTP) for different attributes of a road system. Users’ WTP values are estimated separately for car and bus users. The scope of work includes design of survey instrument, development of behavioural database, analysis of behavioural data using suitable logit model specifications, and estimation of WTP values.

A Stated Preference (choice based) survey was carried out to collect behaviour data from car and bus users. Three alternative scenarios are considered during choice experimentation: (i) An existing undivided two lane road, (ii) A hypothetical undivided two lane improved road, and (iii) A hypothetical Four Lane Road with median. Both quantitative and qualitative attributes are considered in choice experimentation. The data analyzed by developing multinomial logit (MNL) and were random parameter (RPL) logit models. Sparely used constrained triangular distribution of random parameters is attempted for the development of RPL models. The effect of socioeconomic characteristics on the mean of random parameter (called ‘mean heterogeneity’) is investigated with RPL models.
Findings and Conclusions
Users’ WTP values associated with various attributes of travel are estimated separately for car and bus users. A comparison of WTP estimates indicates that utilities associated with speed and comfort are more as compared to the same due to other attributes of road improvement. While personal car users have higher value of WTP for quantitative attribute speed, the bus users have higher value of WTP for qualitative attribute comfort. The work also demonstrates successful application of constrained triangular distribution in the development of RPL models. A comparison of different econometric models clearly shows an improved model performance for RPL models over MNL model. Among RPL models, the model performances improved further when the mean heterogeneity is taken into consideration. The toll rates for car and bus users are estimated by charging 25% of user benefits to respective users.