Exploring Comparative Ridership Drivers of Bus Rapid Transit and Light Rail Transit Routes

A major research gap is the relative ridership performance of Bus Rapid Transit (BRT), Light Rail Transit (LRT), and streetcar (SC). This paper assesses ridership influences of 101 routes in Australia, Europe, and North America using multiple regression examining the influence of transit mode, vehicle capacity, service level, employment/residential density, car ownership, speed, stop spacing, right-of-way, vehicle accessibility, and integrated fares on ridership (boardings/vehicle km; BVK). Average ridership is higher for LRT/SC routes than for BRT routes, and although service levels vary greatly, they are lower on BRT systems. Residential/employment density is higher for LRT/SC routes compared to BRT. A regression model predicting BVK was significant (R2 = 0.83) with six predictors: being in Europe, speed, vehicle capacity, employment density, service level, and integrated ticketing. Results suggest that the transit mode does not directly impact ridership but rather acts through vehicle size and service levels. Limitations and opportunities for future research are identified. Introduction Cities facing the challenge of expanding transit often find themselves weighing the relative costs and merits of Bus Rapid Transit (BRT) versus Light Rail Transit (LRT). However, a major research gap is empirical assessment of the comparative merits of BRT versus LRT. Although the relative costs of LRT and BRT have been analyzed Journal of Public Transportation, Vol. 16, No. 2, 2013 48 (U.S. General Accounting Office 2001; UK Commission for Integrated Transport 2005), there is almost no research that explores relative ridership impacts of one mode over the other. This paper presents the findings of an empirical route level analysis of the factors influencing ridership on a series of BRT, LRT, and streetcar (SC) routes in Australia, Europe, and North America. Its aim is to provide an objective base to determine whether the transit mode has a significant influence on ridership above and beyond the influence of other important variables such as service level or urban density. The research integrates the data sets from two separate studies predicting the ridership of BRT systems in Australia (Currie and Delbosc 2011) and LRT/ SC ridership in Australia, Europe, and North America (Currie, Ahern, and Delbosc 2011). The analysis will help inform cities that are comparing the relative merits of a BRT or LRT system for their needs. The paper is structured as follows. The first section overviews previous research associated with route-level ridership drivers. This is followed by a discussion of the methodology used to collate and analyze the data. Results are then presented, followed by conclusions from the research. Previous Research A summary of previous research on factors that influence LRT and BRT ridership is presented in Table 1. High service levels, measured in terms of frequency and span of hours covered, has often been cited as an important driver of patronage on all public transport modes. Urban density is also identified as an important influence: “Nearly every study that has focused on transit ridership has provided evidence that density is the primary determinant of transit ridership” (Johnson 2003, 32). Much research cites the importance of an integrated public transport network as a key driver of high light rail patronage (FitzRoy and Smith 1998; Denant Boemont and Mills 1999; Babalik-Sutcliffe 2002) and transit patronage in general (Nielsen et al. 2005). Patronage drivers in this case involve service and fare integration as well as the wider “network effects” these can generate. A range of other factors has been suggested that might also influence light rail ridership. Cheap fares were cited in two reports (FitzRoy and Smith 1998; Kain and Liu 1999). A number of researchers cite the importance of a strong policy context as a basis for high light rail ridership (e.g., Knowles 2007). Several researchers have suggested that high car ownership can