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.
This chapter identifies checklists and guidelines that could be used to improve the state of the practice in toll demand and revenue forecasting. Whereas the preceding chapter reviewed practices in specific topics, the checklist and guidelines could provide a framework within which these topics can be addressed. Three examples are taken from the literature. They comprise a checklist, guidelines, and an index: the checklists and guidelines constitute lists of ques- tions and issues that should be addressed in a toll road demand and revenue forecast, whereas the index provides a way to organize and understand the factors and parameters that influence the forecast. All are aimed at helping the facil- ity owner, proponent, and financial analyst (and their mod- elers) to better understand the process and identify questions that should be asked in the development of toll road demand and revenue forecasts. CHECKLISTS AND GUIDELINES FOR MODELS Federal transportation planning legislation requires that each MPO develop a transportation plan as part of its planning process (70). A transportation plan requires forecasts of future demand for transportation services that are usually arrived at by using travel demand models. Of specific inter- est to the toll demand and revenue forecasting community should be the documentation and access to the planning assumptions and forecasting methods used in the travel demand modeling process. FHWA has compiled a checklist for travel demand forecasting methods, mainly with the pur- pose of providing a certification review team with an overview of travel forecasting methods used by MPOs. Spe- cific examples of important planning assumptions included in the checklist are: ⢠Population change (should be compared with past trends and with statewide demographic control totals). ⢠Employment change (should be compared with past trends and with statewide economic growth control totals). ⢠Regional distribution of future population, employ- ment, and land use. ⢠Demographic change (including automobile owner- ship, household income, household size, and multi- worker households). ⢠Travel behavior change (including telecommuting, trip chaining, and Internet shopping). 36 Specific examples of important forecasting methods included in the checklist are: ⢠Last model revision (i.e., when were new variables, new algorithms recalibrated with new data?). ⢠Model specification (i.e., choice of model, specification of key model coefficients). ⢠Calibration data (what data were used; e.g., National Household Travel Survey or Census Transportation Planning Package). ⢠Local survey (how was the survey conducted, what type of control?). ⢠Model validation (what year and data source was the model validated against?). ⢠Size of network. ⢠Number of zones. ⢠Non-home-based travel (e.g., freight commercial ser- vices and tourists). One analyst identified a list of recommended enhance- ments, from the perspective of the financial community, which would make the results more acceptable and more likely to qualify for investment grade status. The recommen- dations included (5): ⢠Incorporation of a range of possible outcomes given the low probability that the base case forecast will exactly match the likely outcome. ⢠Further study and greater validation of value of time as an input in forecasting models. ⢠Further study and greater validation of the ramp-up effect on start-up toll road facilities. ⢠More detailed truck traffic analysis as the higher rev- enue margin created by trucks is an important compo- nent of a forecast, especially when trucks are projected to be a significant fraction of traffic. ⢠Incorporation of the risk and reward of electronic toll collection with respect to violations and toll evasion against faster throughput, ease of use, and revenue recovery through penalties. ⢠Enhance the investors understanding of the exposure to modeling while highlighting risk in the final product (i.e., enhancing the validation process by validating more than one year, full disclosure of model limitations, etc.). In a review of the performance of 14 toll road projects, another financial analyst identified the key variables and CHAPTER FOUR CHECKLISTS AND GUIDELINES TO IMPROVE PRACTICE
37 inputs that he believed have had large repercussions, both positively and negatively, on toll road demand and revenue forecasts (6). One key variable was economic activity. Although national economic trends were relevant, the economic activ- ity within the region and project corridor had a greater impact. An example of this is in Harris County, Texas, where a sharp downturn in oil prices in 1986 left economic growth in the region well below the projected levels. Particularly hard hit was the downtown business district, the primary end destination for many vehicles that were projected to use the toll road, which caused the forecasted traffic demand to be considerably lower than was projected. An input that was constant in all the successful forecasts was the use of 30% or less as projected revenue growth over the first 5 years of operation. This low projection explicitly captured the high initial demand of the road with no need for toll increases. In contrast, where a very high revenue growth rate was assumed at the start of the proj- ect, it indicated a dependency on the growth of these routes to meet the forecasted revenue. If these growth rates were not met, the lost revenues were not easily recaptured in the following years; indeed, the forecast continued to lag behind initial projections. Other inputs and variables that appeared most crucial in the model forecasting process were time saved, the cost of travel, and the ability or willingness to pay of potential users. In summary, the most successful forecasts generally had accurate or even conservative economic forecasts with mod- erate anticipated growth levels. These toll roads were built in corridors that were fully developed and where congestion already existed. Another factor was that the corridor income levels were above the regional levels and time savings were in excess of 5 min. GUIDELINES SPECIFIC TO TOLL ROAD FEASIBILITY STUDIES The Texas Turnpike Authority (TTA), a division of the Texas DOT, provides an example of recommended guide- lines for conducting traffic and revenue studies in support of toll feasibility analyses (73). The key goals of the guide- lines were to outline the traffic and revenue reporting requirements, improve the consistency of assumptions, and improve quality assurance. Four levels of analysis were proposed (74): ⢠Conceptualâdetermines the potential for a toll road project to support bonds. (Expected durations of each type of study were provided and they are listed here as an indication of the level of effort and detail. The con- ceptual level had an estimated duration of 1â4 weeks.) ⢠Sketchâproject-specific estimate of costs, demand, and revenues (6 weeks duration). ⢠Intermediateârefines the previous analysis, including a tolling plan. It is expected that demand projects would be derived from a travel demand model (4â6 months duration). ⢠Investment gradeââextensive and detailedâ analysis âto determine its value in anticipation of proceeding to the bond marketâ (12â18 months duration). The guidelines recommended a forecast period of 40 years beyond the opening date of the project. The following inputs should be taken into account (73): ⢠Average daily tolled and non-tolled volume, ⢠Weekday toll transactions, ⢠Gross annual revenue, ⢠Tolled length, ⢠Number of lanes being tolled, ⢠Truck percentage, ⢠Opening year automobile and truck toll rate, ⢠Toll increase increment, ⢠Period between toll increase, and ⢠Equivalent revenue days. The guidelines noted that a lack of consistency on key assumptions affected the comparability of options. Accord- ingly, to help ensure consistency and improve comparability, the following parameters should be considered (73): ⢠Phased improvement or system implementation scenar- ios; that is, each tolling point should change in response to the addition of new tolled segments. ⢠The definition of the study area should encompass those transportation facilities that could influence the candidate toll project. Route classification and lane configurations of competing facilities should also be considered. ⢠Toll diversion assumptions, including general toll attraction and composite toll attraction. Composite toll attraction includes ramp-up, electronic toll collection, toll rate adjustments, and toll utilization. ⢠Toll transactions estimates are required for all four of the study levels described previously. ⢠Traffic growth constraints should be considered, owing to the long time frame (40 years) of the forecast. Con- straints could include highway corridor capacity, com- peting toll facilities, economic development, etc. ⢠Trip rate equivalence or toll equity, which considers the average toll rate paid by the user traveling all possible originâdestination paths on the facility. The toll rates should fall within an acceptable toll rate ratio. The TTA has prepared a series of brochures related to tolling, including a âToll Feasibility Analysis Guide,â which summarized the four levels of analytical studies and their main characteristics (74).
38 Another TTA document noted that the major bond rating agencies looked for the following topics to be addressed as part of toll road demand and revenue forecasts (2): ⢠Land use and demographic assumptions, including pop- ulation and employment information; ⢠Highway network and alternative routes that are both feeding the project or competing with the project; ⢠Weekday versus weekend traffic; ⢠Review of travel demand parameter assumptions; ⢠Trip-making characteristics; ⢠Truck percentage and generated revenue, because of the impact trucks can have on toll revenue; ⢠Peak-period versus off-peak management, especially in managed lane or congestion pricing projects; ⢠Value of time; ⢠Ramp-up period; ⢠Violation rate; ⢠Toll rates and increases; ⢠Point estimate forecasts; and ⢠Economic and political risk. TRAFFIC RISK INDEX One financial analyst developed a Traffic Risk Index to assess and compare the risk associated with a given traffic and revenue forecast according to 10 facility attributes. Most of the attributes are divided into subattributes. A notional scale of from 0 to 10 assessed the risk for each attribute, with higher values reflecting increasing inherent uncertainty. Descriptions are provided for each extreme to help illustrate the range of risk that would be considered for the particular subattribute. For example, the fourth attribute, âforecast horizon,â refers to near-term forecasts as having the least degree of uncertainty and long-term (â30+ yearâ) forecasts as having the greatest uncertainty. The Index was described as a âstarting point for consider- ing toll-project traffic uncertainty in a logical and consis- tent manner. The Index also represents a checklist that can be used to examine project-specific uncertainties and prompt appropriate enquiries (allowing the analyst to draw his or her own conclusions about the likely reliability of the resultant forecasts)â (9). The means of assessing the risk are determined by the user, who also can tabulate or weight the results at his or her discretion. The Index is reproduced in Table 3. SUMMARY OBSERVATIONS Four observations are useful in summarizing the checklists, guidelines, and indexes. ⢠Each attempts to introduce consistency and a system- atic approach to developing traffic and revenue fore- casts. ⢠However, neither method nor the application of the lists, guidelines, or indexes is specified or prescribed; rather, these are left to the discretion of the user. ⢠A large range of attributes is described. ⢠The types of attributes that are considered varies according to the perspective: The user (i.e., financial) communityâs lists and indexes generally describe the conditions and environment within which the traffic model and forecasts are developed, as well as on the assumptions and inputs, whereas the modeling commu- nity focuses on the model specification methods.
39 Project Attribute 0 1 2 3 4 5 6 7 8 9 10 Tolling regime ⢠Shadow tolls ⢠User-paid tolls Tolling culture ⢠Toll roads well established-âdata on actual use are available ⢠No toll roads in the country, uncertainty over toll acceptance Tariff escalation ⢠Flexible rate setting/escalation formula; no government approval ⢠All tariff hikes require regulatory approval Forecast horizon ⢠Near-term forecasts requirement ⢠Long-term (30+ year) forecasts Toll facility details ⢠Facility already open ⢠Facility at the very earliest stages of planning ⢠Estuarial crossings ⢠Dense, urban networks ⢠Radial corridors into urban areas ⢠Ring roads/beltways around urban areas ⢠Extension of existing road ⢠Greenfield site ⢠Alignment: strong rationale (including. tolling points and intersections) ⢠Confused/unclear road objectives (not where people want to go) ⢠Alignment: strong economics ⢠Alignment: strong politics ⢠Clear understanding of future highway network ⢠Many options for network extensions exist ⢠Stand-alone (single) facility ⢠Reliance on other, proposed highway improvements ⢠Highly congested corridor ⢠Limited/no congestion ⢠Few competing roads ⢠Many alternative routes ⢠Clear competitive advantage ⢠Weak competitive advantage ⢠Only highway competition ⢠Multi-modal competition ⢠Good, high-capacity connectors ⢠Hurry up and wait Surveys/ data collection ⢠âActiveâ competition protection (e.g., traffic calming, truck bans) ⢠Autonomous authorities can do what they want ⢠Easy to collect (laws exist) ⢠Difficult/dangerous to collect ⢠Experienced surveyors ⢠No culture of data collection ⢠Up to date ⢠Historical information ⢠Locally calibrated parameters ⢠Parameters imported from elsewhere (another country?) ⢠Existing zone framework (widely used) ⢠Develop zone framework from scratch Users: private ⢠Clear market segment(s) ⢠Unclear market segments ⢠Few key origins and destinations ⢠Multiple origins and destinations ⢠Dominated by single-journey purpose (e.g., commute, airport) ⢠Multiple-journey purposes ⢠High-income, time-sensitive market ⢠Average/low-income market ⢠Tolls in line with existing facilities ⢠Tolls higher than the norm extended ramp-up? ⢠Simple toll structure ⢠Complex toll structure (local discounts, frequent users, variable pricing, etc.) ⢠Flat demand profile (time-of-day, day- of-week, etc.) ⢠Highly seasonal and/or peak demand profile Users: commercial ⢠Fleet operator pays toll ⢠Ownerâdriver pays toll ⢠Clear time and operating cost savings ⢠Unclear competitive advantage ⢠Simple route choice decision making ⢠Complicated route choice decision making ⢠Strong compliance with weight restrictions ⢠Overloading of trucks is commonplace Micro- economics ⢠Strong, stable, diversified local economy ⢠Weak/transitioning local/national economy ⢠Strict land use planning regime ⢠Weak planning controls/enforcement ⢠Stable, predictable population growth ⢠Population forecast dependent on many exogenous factors Traffic growth ⢠Driven by/correlated with existing, established, and predictable factors ⢠Reliance on future factors, new developments, structural changes, etc. ⢠High car ownership ⢠Low/growing car ownership Source: Bain and Wilkins (9). TABLE 3 TRAFFIC RISK INDEX