Estimating Toll Road Demand and Revenue (2007)

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.

51 OVERVIEW OF METHOD A survey was conducted regarding the practice of toll facil- ity demand revenue estimates. This survey of practitioners was sent to four specific agency types: state departments of transportation (DOTs), toll authorities, bond rating agencies, and bond insurance agencies. The first two types represent the traditional developers and users of traffic forecasts. The inclusion of the financial community (the latter two types) was intended to capture the viewpoints and experiences of the forecasting and modeling process from as many partici- pants involved in the process as possible. Participants were given the option of answering directly online through a web-based survey program (Websurveyor), or completing a hardcopy of the survey, which was included in the e-mail as a pdf (in which the survey form could be returned to the consultant by mail or fax). SURVEY OF PRACTICE Survey Development The survey was conducted over a two-month period in April and May 2005. Before the survey was distributed, it was nec- essary to establish a comprehensive outline and to test the survey to ensure that it was as accurate and target-specific as possible. During the initial stages of the survey design, it was deter- mined that the survey should be completed by several differ- ent types of agencies that are involved in the toll road demand and revenue estimating process. A logical way to construct the survey was to structure the questions as close- ended as possible, to provide a clear indication of what was being asked and to help in the summary process. Therefore, most of the questions were asked with a selection or list of answers and the respondent was instructed to select a single answer, or as many answers as applied, depending on the question. However, given the nature of the subject, a stan- dard set of answers likely would not always have covered all possibilities. Therefore, for every question there was also the option of selecting “Other,” with space provided to allow expansion of the respondent’s answer. The research team developed a survey focusing on state- of-the-practice applications and techniques. Specifically, important information that was targeted for acquisition was: • Who is the respondent to the survey? • What types of modeling processes are being used? • What variables are being modeled? • How are the models being calibrated and validated? • Are there specific case studies or examples? • What is the respondent’s experience with the model forecasts? The survey was divided into three self-contained sections (designated as Parts I, II, and III). This made the survey more “respondent friendly,” to specifically target areas of interest in the modeling and forecasting processes. Part I determined the type of agency that was responding to the survey and who from that agency was responding. It also requested a discussion of their philosophy, in terms of their use of tolling technologies, what type of facilities was tolled, etc. Part II pertained to the forecasting model itself and its variables. This section was answered either by the original respondent, if the responding organization performed the modeling and forecasting in-house, or it could be forwarded to a consultant or the agency that actually developed or applied the model for the original responding agency. This section asked the respondent to describe in detail the type of model used and the parameters of the model in terms of inputs, structure, modeled trip purpose, calibration tech- niques, validation checks, etc. Part III asked the respondent to discuss a specific exam- ple of a toll road traffic demand and revenue study carried out by the responding organization. Again, if the original respon- dent did not do the actual analysis, the survey could be passed to the consultant or outside agency responsible for the study. The purpose of the final section was to determine the results of the previously described model (Part II) and whether there were major or minor problems with the analy- sis, whether they were identified, how and if they were cor- rected, etc. A pilot survey session with one respondent was conducted to assess the usability and readability of the survey before distribution to all participants. The pilot participant remotely completed the survey while speak- ing his/her thoughts out loud over the telephone to the research team. Issues were identified related to the clarity of the survey and its ability to extract the desired informa- tion. Minor modifications were made to the survey fol- lowing the pilot survey session. Subsequently, the survey was sent to state DOTs, tolling authorities, bond rating agencies, and bond insurance agencies throughout the United States. APPENDIX A Development and Administration of Survey

Contact names and e-mail addresses of the DOTs were provided by TRB, for the toll authorities by the International Bridge, Tunnel, and Tolling Association (IBTTA), and for the bond rating agencies and bond insurance agencies through the Topic Panel and other individuals. In accor- dance with the scope of the synthesis, the survey was directed to respondents in the United States, although it was also sent to three Canadian provincial/territorial ministries of transportation and one Canadian tolling agency that had been included in the IBTTA list. Two follow-up reminders were sent at regular intervals to those who had not yet responded. Two methods of completing the survey were provided for the practitioner. The first was to complete the survey online through a commercial online survey host (Websurveyor). The second option was to allow the practi- tioner to complete the survey by hand on a hard copy. To this end, a pdf version was included in the initial e-mail. The survey could be returned by mail or fax. Support was pro- vided throughout the process by means of e-mail correspon- dence, as well as a toll-free telephone number distributed with the survey. Upon review of the survey responses, follow-up tele- phone interviews were initiated with respondents where additional information was required or to clarify unclear or contradictory responses. Moreover, if the survey indi- cated an exceptional toll modeling practice, then further details were sought to document the techniques and methodology. Discussion of Survey Response Two unexpected problems were encountered. These consti- tute important findings in their own right; however, they also affected the results of the survey and the subsequent analysis. Several respondents indicated that they used relatively straightforward spreadsheet programs to generate forecasts. It should be observed that these practices of modeling were used to forecast short-term travel demand and revenue for 52 existing toll facilities that have been in operation for many years. Simple modeling techniques were designed to forecast short-term annual demand and revenue for operations and maintenance, based largely as an extrapolation of previous years’ experience, or by using elasticity factors. There was also an issue pertaining to the compatibility of Parts II and III of the survey. In Part II, respondents were asked to describe the application of their model and the data to a specific demand and revenue forecast of their choice. This focus on a single model, application, and forecast was intended to avoid confusion among those respondents that had several models, applications, or forecasts. Most respondents selected the newest model and its use in a recent toll road feasibility study, generally ensuring that only the latest modeling tech- niques (the state of the practice) were described. However, in Part III, the respondents were asked for their experiences with actual toll road demand and revenue; specifically, how it related to the model described in Part II. Because many respondents had chosen their newest project for Part II, they were unable to provide corresponding state-of-the-practice responses to Part III. This raised an interesting point in general, namely that the state of the practice in toll road demand and revenue fore- casting is difficult to capture definitively, because the latest methods are still unproven or at least there are few actual per- formance data to substantiate whether the new modeling techniques are more or less successful than previous tech- niques. Because of the required lengthy lead time before the completion of the design and construction of toll roads, the only models with enough data to be able to determine their successes and failures are 10 or more years old. Accordingly, a caveat must be placed on what actually constitutes the state of the practice insofar as the results of the modeling process itself are concerned, because some of the new techniques are still in their infancy and the old techniques are being changed and updated. Of the 138 surveys that were sent, there were 55 respon- dents, for a response rate of 40%. The corresponding rates of return by agency are found in Table A1. Of the 55 surveys Agencies No. of Surveys Sent No. of Surveys Returned Rate of Return (%) State DOTs 56 29 52 Tolling authorities 70 21 30 Canadian authorities 4 3 75 Bond rating agencies 3 2 67 Bond insurance agencies 5 0 0 Total 138 55 40 TABLE A1 RATE OF RETURN BY AGENCY

53 returned, 10 were e-mail responses indicating that the state did not own or operate, or was not planning to own or oper- ate, a toll road. Question I.1 of the survey asked respondents to indicate their interest or mandate in toll road facilities by selecting one of the following five options: • We are currently or plan to be an owner/operator • Bond rater/insurer • Travel forecasting or other consultant/independent reviewer • None planned or expected • Other. Of the remaining 45 survey respondents, 19 indicated that they were neither planning nor expecting to own or operate a toll road, and therefore did not complete the remainder of the survey. Of the remaining 26 respondents, three selected “other” (see Appendix B for detailed responses) and one of those respondents indicated that he/she could not answer the survey. In summary, there were a total of 25 completed responses to the survey. Of these, 23 completed Part I (two bond rating agencies that responded were directed to skip Part I, and also skipped Part II), 13 respondents completed Part II, and 13 respondents completed Part III. Tables A2 and A3 sum- marize in more detail which types of agencies completed Part II and Part III, respectively. Of the total number of respondents, only 23 (42%) indi- cated that they were currently or planning to be an owner or operator of a toll road facility. Table A4 shows the percent- age for all respondents. A complete summary of the survey results is provided in Appendix C. Agencies No. of Respondents That Currently or Plan to Be an Owner/Operator No. of Survey Respondents Percentage State DOTs 7 29 24 Tolling authorities 15 21 71 Canadian authorities 1 3 33 Bond rating agencies 0 2 0 Bond insurance agencies 0 0 0 Agencies Completed Part II State DOTs 3 Tolling authorities 9 Canadian authorities 1 Bond rating agencies 0 Bond insurance agencies 0 Total 13 Agencies Competed Part III State DOTs 2 Tolling authorities 8 Canadian authorities 1 Bond rating agencies 2 Bond insurance agencies 0 Total 13 TABLE A2 RESPONDENTS COMPLETING PART II TABLE A3 RESPONDENTS COMPLETING PART III TABLE A4 SUMMARY OF RESPONSES TO QUESTION I.1 OF SURVEY