Modeling Generalized Cost of Travel for Rural Bus Users

In order to formulate rational improvement proposal for rural bus services, it is necessary to understand how people value different attributes of travel. In this article, the disutilities of travel have been modeled based on stated choice data collected from trip-makers traveling along a rural bus route in Midnapur district, West Bengal, India. Multinomial Logit Model (MNL) is used to develop utility equations and the total disutility of travel is estimated in the form of generalized cost. The perceived values associated with in-vehicle travel time, service headway, and comfort level for the study route are estimated and found to be significant. Introduction Passenger transportation demand in rural India is largely served by the bus transportation system. In India, more than 70 percent of the population is located in rural and suburban areas. The rural population predominantly consists of lowincome households with very low car ownership. Therefore, the rural population is almost completely dependent on the available bus transportation system, creatJournal of Public Transportation, Vol. 7, No. 2, 2004 60 ing a vital role for rural bus service in the economic growth and development of the country. Despite ample opportunities for improvement of rural bus transportation in India over several decades, this mode has not drawn adequate attention from transportation professionals and policy-makers. Improvement may be in the form of increase in frequency, comfort level, travel speed etc. However, every possible improvement in existing service is likely to be attributed to an increase in fare level. Therefore, for judicious improvement planning, it is necessary to understand users’ perception about various attributes of travel. The objective of this article is to understand rural bus users’ perceptions for different attributes of service and model the generalized cost of travel. A typical rural route served by the bus transportation system in India has been considered in the case study. Many researchers have attempted to model people’s perceptions about various attributes using Revealed Preference (RP) and/or Stated Preference (SP) data (Adamowicz et al. 1994; Bates 1982; Kroes and Sheldon 1988; Louviere 1988; Hensher 1994; Jose Holguin-Veras 2002). RP requires a large sample size and cannot accommodate hypothetical alternatives. SP surveys gained importance over RP due to their smaller sample size requirement and their ability to accommodate hypothetical alternatives yet produce results comparable to/on par with RP results (Hunt 2001). Multinomial Logit (MNL) modeling has been widely accepted by researchers and practitioners for analyzing the RP or SP data (Louviere and Woodworth 1983; Jose Holguin-Veras 2002; Hunt 2001). Attributes considered in utility equations, developed by MNL model, have different measuring units. Conversion of these attributes into a common unit enables comparison or estimation of relative importance of each attribute over the other. Summation of these converted attributes is called the generalized cost. In this article, modeling of generalized cost of travel has been demonstrated with reference to a rural bus route in India, which is connecting a district head quarter (Midnapur) and a tourist place (Digha) in West Bengal, India. The two areas are connected by a direct bus route of 142km. Travel demand along this route is largely served by ordinary bus service. The bus service takes about 5 hours to cover the distance of 142km and serves about 35 intermediate stops. For the development of utility model, it is necessary to create a database with the help of SP and/or RP observations. Normally, pure SP-based data should be avoided for the development of discrete choice models, as the reliability of parameter Modeling Generalized Cost of Travel for Rural Bus Users 61 estimates could be low. However, for the study bus route all users are essentially captive riders and do not have a choice other than accepting the existing bus service. Therefore, for the development of utility model for rural bus users, the database is created based on only SP observations. Further, there is currently limited information in the literature on stated choice experiments in the context of rural passenger transportation in India or other developing countries. Therefore, before creating a large database for the development of a refined model, a limited number of observations were obtained and analyzed. The utility model developed based on these observations is reported in this article.