Toolkit for Estimating Demand for Rural Intercity Bus Services (2011)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

S U M M A R Y Purpose of Project The objective of this research project was to develop a sketch-planning guide and sup- porting tools that could be used by state transportation department program managers and both public and private rural intercity bus service providers to forecast demand for rural intercity bus services. The research approach documented in this report accomplished this objective by: • Conducting stakeholder interviews of federal and state officials, industry and professional associations, key intercity carriers, consultants, and others to determine the current state of demand forecasting and to identify examples of existing rural intercity services that were contacted to obtain service descriptions and ridership data. • Conducting surveys of rural intercity projects to get details on the nature of the proj- ects, including service characteristics, service area, ticketing and information, ridership (including trends), and forecasting methods used to plan the services. • Identifying and evaluating existing rural intercity bus forecasting methods from infor- mation supplied by stakeholders, providers, and the literature. • Developing a sketch-planning guide and supporting tools, data, and methodologies to enable users to forecast rural intercity bus ridership. Initially provided as a conceptual framework, these products were revised and refined following input from the TCRP Project B-37 panel to result in a user-friendly final product. • Providing a final report and Microsoft® PowerPoint presentation to document the research process and the forecasting tools and to provide a presentation that was used by TCRP, the study team, and panel members to describe the research and tools. The potential audience for this research includes state agency program officials and staff, planners, local officials, and existing and potential public and private operators and sponsors of rural intercity bus service. Background The national intercity bus network has been contracting in coverage for many years, but a substantial shift away from services in rural areas began with the passage of the Bus Regulatory Reform Act in 1982. Following the loss of substantial amounts of rural intercity bus service subsequent to regulatory reform, the Intermodal Surface Transportation Efficiency Act (ISTEA) passed by Congress in 1991 created the Federal Transit Act Section 18(i) pro- gram of assistance for rural intercity services, offering operating, capital, and administrative Toolkit for Estimating Demand for Rural Intercity Bus Services 1

2funding to the states for use in maintaining or developing rural intercity services. This program subsequently codified as Section 5311(f) of Title 49. The availability of this funding and the state-funded programs in several states calls for a tool to identify which potential rural intercity feeder markets make sense, based on the projected ridership and revenue. Currently there is no demand model, rule of thumb, or similar tool that is based on recent experiences to assist in determining the likely intercity- related ridership, and the impact of different arrangements on the potential demand. Most basically, a way to estimate intercity trip demand from rural areas to larger cities is needed to help in the design of projects that will link rural areas with major urban areas and the national intercity network. The level of demand obviously varies with population, and prob- ably with frequency and service design, and is a major consideration in service design issues. Review of Demand Estimation Methods Chapter 2 of this report documents a number of approaches to the estimation of rural inter- city bus demand. During the 1980s as the bus industry restructured following deregulation, the interest in potential state or federal programs to provide operating or capital assistance led to a number of efforts to develop demand models. More recent efforts at planning have used earlier models, or other sketch-planning techniques, to estimate potential ridership. The approaches used in the various studies have varied according to the desired application and the available data. Approaches have included the use of: • Per capita intercity trip generation rates • Ridership on comparable services • Historical data • Stop-level regression models • Route-level regression models • City-pair regression models • Network models Several applications of these approaches are documented, including the use of trip rates in the Washington intercity bus plan and the use of a regression model to estimate demand and revenue for a Virginia study. Inventory of Existing Rural Intercity Routes and Ridership An important and significant part of the effort to develop a demand model for rural intercity bus ridership involved an effort to identify current or recent rural intercity bus services, their characteristics, and their ridership. These basic data elements are critical to the ability to calibrate or evaluate any type of technique for estimating ridership. Chapter 3 describes the type of data sought and the survey methodology to collect the data. Initially all the service characteristics that could potentially affect ridership were identified, and the list used to develop a survey for completion by the agency or firm operating the service. Initial pilot tests of the survey resulted in a shortened version. A second step involved the identification of rural intercity services. Because it was antic- ipated that the resulting models would be used primarily to estimate ridership on services funded with Section 5311(f) operating assistance, the approach taken involved contacting the transit programs in all of the state departments of transportation to determine if they had provided operating funding for rural intercity bus service in the past three years. If so, the study team requested contact information for the provider and any information available

at the state level on service characteristics or ridership. Additional effort went into using other data sources such as websites and industry schedule guides to develop service characteristics. The effort involved in identifying the state contacts, contacting carriers, and obtaining service and ridership data was significant. The result was a database of routes, with data on the operator, route endpoints, stops, route length, frequency, fare (and/or fare per mile), corridor population, destination population, and the presence of key generators (college or university, major medical center, airport, etc.). Efforts to include more detailed demographic information were initiated but were difficult to apply on a corridor or route level. However, the result of the overall survey effort produced enough route-level data that creation of some kind of tool for estimating rural intercity demand was deemed possible. A total of 133 routes were identified. Rural Intercity Bus Classification Scheme With the database of routes and route characteristics in hand, the study team developed a classification of the services in an effort to combine services with similar characteristics into different classes as a means of possibly developing separate demand tools and clustering similar services to identify potential commonalities to assist in developing demand estimation techniques. This process is described in more detail in Chapter 4. An initial classification was developed based on the type of provider. Three classes were developed: • Services that are comparable to traditional intercity bus services • Services that are regional in character, provided by private firms • Services that are regional, but are operated by public transit providers For each class the characteristics of that class were identified, based on the data available in the database from the survey. At this point the interim report, including the data and classification, was presented to the TCRP Project B-37 panel. The panel expressed concerns that many of the services were not truly “intercity” in nature and that use of the Section 5311(f) definition of intercity service as a criteria for inclusion in the study allowed many regional or rural transit services with widely varying characteristics to be included in the database. The database, including classifications, was intended to be the basis for calibrating any kind of model or tool. With the assistance of one of the TCRP B-37 panel members, a reclassification of all routes in the database followed. Because the revised criteria focused on connectivity to the national intercity bus network as a key element of the definition of intercity service, additional data was needed on many of the services to determine if a passenger could use the service (included in the database) to access the national intercity bus network. Obtaining this data involved checking for service to common or nearby terminals, reviewing schedules, and, in a number of cases, trying to determine if any passengers on a given route actually boarded or alighted at an intercity terminal that happened to be located on a route. In addition, an effort was made to identify and include rural intercity routes that did not receive Section 5311(f) funding. Although several routes were identified, most carriers did not provide data for those routes as they are not required to report ridership to any public entity on such services. The revised classification included 99 routes, all considered “rural intercity” for the purposes of this study. The study team cannot provide any comparison of unsubsidized rural intercity routes with Section 5311(f) subsidized services that would support conclusions regarding the extent to which the model results may underestimate ridership as a result of being calibrated with data from subsidized routes. 3

4Development of a Sketch-Planning Tool The process of developing demand estimation tools, even with a fairly large data set, proved to be more problematic than originally thought. Chapter 5 describes the efforts made and many of the issues that developed during this process. Initially the study team considered all of the desired characteristics of a model or toolkit. This process helped to set the goals for the effort, but also made it apparent how difficult it might be to address all the potential issues that might be faced by a user. This project was intended only to develop a demand estimation tool or process, not to develop a full plan- ning process—yet full use of a tool or technique might well require a great deal of additional background or education for a user to be able to be sure that this technique would be appro- priate, to obtain data, and to interpret results. Two basic development approaches were undertaken. One involved the effort to develop trip rates for the routes and corridors included in the database, potentially including route length as a factor to adjust trip rates. However, when no discernable pattern of trip rates could be developed, two issues were identified. One is the impact of intermediate stops on route-level ridership, and the other is the difficulty in determining the appropriate corridor population to calculate a trip rate when a large metropolitan area is part of the corridor. In such cases a trip rate that includes the large population will be very different from a route with only rural stops. Eventually it was decided to see if trip rates from a separate source could be used to develop a tool that would have predictive value. A special run of the National Household Travel Survey (NHTS) focusing on the long-distance trips made by persons in non-urbanized areas was requested. The resulting data was also classified by income and by region. Information on mode share from several sources was used to develop trip rates—the 1 percent mode share produced ridership estimates most similar to the survey data, and it was chosen for use in the trip rate model or tool. The alternative development approach taken was an effort to develop a multiple-regression model using the database data. Initial efforts produced models with limited explanatory power. Evaluation of these initial results led to a disaggregation of the population data variable, which was corridor population, into urbanized and non-urbanized components. Finally improved results came from using populations for urbanized areas (over 50,000 persons), urban clusters (2,500 to 50,000 persons), and Census-designated places (under 2,500 persons). These groupings provide populations that are not necessarily limited to municipal boundaries. Analysis of residuals led to continued work with the regression model, this time reducing the cases to eliminate routes that were outliers. A separate variable for the number of stops was also included in the data set. With the elimination of outliers, the data set was reduced to 58 usable cases, and the distinction between standard intercity bus service and regional rural intercity bus service classes was made into a categorical variable. Continued work with stepwise regression eventually resulted in the best fitting model: R Adjusted R a2 20 712 0 690= =. , . Ridership average origin popul= − +2803 536 0 194. . ation the number of stops on the route ( )+ ( 314 734. )+ ( ) + 4971 668 578 . airport service or connection 3 653. service provided by intercity provider( ) aIn a regression equation, the term “R2” refers to the fraction of the sample variance of the dependent variable that is explained by the regressors. “Adjusted R2” is a modified version of R2 that does not necessarily increase when a new regressor is added to that regression. In general, a higher value of R2 means that the model has more explanatory power. See pp. 193–195 in Introduction to Econometrics, James H. Stock and Mark W. Watson, 3rd Edition, Pearson Education, Boston.

Where: Ridership = annual one-way passenger boardings Average origin population = sum of the populations of origin points (all points on the route except that with the largest population) Number of stops = count of points listed in public timetables as stops Airport service or connection = route serves an airport with commercial service either directly or with one transfer at a common location Intercity provider = service operated by a carrier meeting the definition of an intercity bus carrier (see Definition of Intercity Bus Service in Chapter 6) A subsequent effort used the residualsb from the regression model to adjust the trip rate model results, improving the results slightly over the pure trip rate model as shown in Table S-1. Both of these techniques are more accurate for current rural intercity bus services than demand models developed for NCHRP in 1980. They represent a pragmatic approach that makes use of available data to produce initial estimates of potential ridership for new rural services. The regression model has the correct signs (e.g., ridership increases with a higher population base, etc.) and is plausible given general knowledge about travel behavior. It reflects higher ridership for intermodal connectivity to airports and for interlining. It utilizes popu- lation data as a key variable, but the impact of population is moderated by using the number of stops to calculate an average population per stop. This scenario is plausible in that the study team expects ridership to be lower if the bus stops a lot to serve that population, which seems to reflect market preference for fewer stops. The use of the NHTS trip rate data also involves making maximum use of the available data. It provides ridership estimates based entirely on population served, but it is calibrated in a sense through the selection of the mode choice factor to provide ridership estimates that most closely match the usage found in the data set. Regional variation is introduced through the use of regional trip rates. Finally, the 58-route data set was used to develop an adjustment factor that can be applied to the trip rate model results to further improve its results. The result is that the trip rate model and the regression model have comparable accuracy in terms of the percentage of time they will predict a ridership figure that is within a given percentage of the actual. However, they may not give the same answer. Both the difficulties experienced and the results suggest that, over the past 30 years, rural intercity bus service has become much more specialized, with the remaining routes or services much more likely to be provided in areas with fairly unique demand characteristics. Neither model takes account of the overhead traffic (ridership originating in or destined to places beyond the endpoints of the particular route in question) that might result in ridership variance or other variables, such as the presence of a large university or military base, that might affect demand. 5 Regression Predictions Within 50% of actual ridership Within 10% of actual ridership Within 5% of actual ridership 1% Trip Rate Prediction 45.60% 14.00% 8.80% Adj. 1% Trip Rate Prediction 54.40% 15.80% 5.30% 59.60% 17.50% 5.30% Table S-1. Accuracy of trip rate and regression models. bIntroduction to Econometrics, James H. Stock and Mark W. Watson, 3rd Edition, Pearson Education, Boston, pp. 190–191.

6The Toolkit The major product of this project was intended to be an easy-to-use toolkit to assist planners in estimating ridership on rural intercity routes. It was decided that the tools would best be provided on a CD with the models and their calculations embedded so that users would not have to deal with formulas or look up tables—but would merely need to input data for a proposed route to get the model estimates. Users desiring more information about the models and the data can read this technical report. Given the degree to which both models depend on population data, the decision was made to include all this data on the same CD as the models, so that the user could designate the stops on a potential route and at the same time obtain the populations and apply them in the models. With the data and the models on the CD, it seemed logical to include the instructions, qualifications, adjustments, examples, and peer data all on the same disk, and set it up so that it would provide links to this additional information at the appropriate places. The toolkit is thus a disk, and the only written directions on the disk involve the type of software needed and how to open the software (this information is also available in Appendix E). Once the toolkit is opened, it provides the user with a discussion of its applicability, an overview of the elements included, a step-by-step process for estimating ridership (which includes preliminary aspects that would precede use of the models and the information that will be needed from the user), possible manual adjustments to improve accuracy, a detailed example of its application to a case, and a lookup database that provides ridership on comparable routes and a link to more descriptive data about the comparable routes. Conclusions Finally, the conclusions about the process and the model include an assessment of the reasons for the difficulty in coming up with a predictive model, the limitations of the two approaches used, and identification of future research needs. The two models developed in this process are limited in that they are not sensitive to changes in fares or frequency and they do not account for ridership that might arise from a population not directly served by the route—for example, through passengers who use the service because it bridges two other routes or riders coming from other modes or going to places with no population (parks, for example). The trip rate model relies on data from the previous NHTS, and the population data is from the 2000 Census, so an update may be needed within a year or two. Future research on intercity bus demand could include additional effort to obtain data on more routes, particularly as the Section 5311(f) program expands. Models to predict demand at a stop would also be useful, as would tools that could allow planners to gauge the impacts of higher frequencies or lower fares. The impact of the availability of long-term parking at stops or terminals is another factor that could be considered in future research. Finally, a major step in developing a tool for estimating intercity bus demand generally would be a network model that would allow for the inclusion of overhead ridership, facilitating the estimation of demand for service to fill network gaps as well as serve populations on a route.