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.

81 APPENDIX C Summary of Survey Responses Part I. Background information I.1 What is your organization’s interest and/or mandate in toll road facilities? Please check (✓) one box only. Responses* We are currently or plan to be an owner/operator 21 Bond rater/insurer 2 Travel forecasting or other consultant/independent reviewer 0 None planned or expected 19 Other (explanation below): 3 *Note: There were also 10 agencies who responded by e-mail citing various reasons why they were unable to complete the survey at the time it was sent. A total of 45 agencies submitted the survey. Comments related to the three respondents who selected “Other”: 1. [A state DOT] is planning a feasibility study 2. [A responding DOT] proposed merger with [a tolling authority] 3. [A state DOT] is working to determine the feasibility of tolling, but there is not have enough detail to answer these questions I.2 What types of tolled facilities does your organization currently or plan to own or operate? Please check (✓) all boxes that apply. Facility Type Currently Own Currently Operate Plan to Own Plan to Operate Urban expressway 8 7 11 11 Rural or intercity expressway 2 2 6 5 HOT lane 0 0 2 2 Express toll lane 2 1 1 1 Bridge or tunnel 8 8 6 4 Urban arterials 0 0 2 2 Rural arterial highways 1 1 2 2 Other (explanation below): 3

82 Comments related to the three respondents who selected “Other”: 1. Currently Own—Toll supported bridges 2. Not Stated—[A tolling authority] currently operates 90 centerline-meters 3. Not Stated—3 toll bridges, including 2 at the border I.3 What tolling technologies are used or are planned? Please check (✓) all boxes that apply. Tolling Technologies Currently Use Plan to Use Traditional toll plaza (manual toll collection only) 5 1 Toll plaza with combination of manual and electronic toll collection (EZPass, FasTrak, etc.) 13 10 Fully electronic toll collection (transponder, record license plate) 2 7 Other (explanation below): 4 Comments related to the four respondents who selected “Other”: 1. Not Stated—Study only at this time 2. Not Stated—We also have coin machines 3. Not Stated—High-speed electronic toll collection lanes are used 4. Plan to Use—Open road tolling I.4 What sources of financing were used or are planned for the construction of these toll road facilities? Please check (✓) all boxes that apply. Source of Financing Currently Use Plan to Use Federal government (all sources) 7 9 State government 7 11 Local/county/district government 4 5 Public–private partnership 4 7 Private sector 1 0 Bond financing 15 11 Other (explanation below): 8 Comments related to the eight respondents who selected “Other”: 1. Plan to Use—Any funding sources will be explored 2. Currently Used—Development impact fees 3. Not Stated—State funds for some [rights-of-way] and [for maintenance] 4. Not Stated—Project conceived as public–private partnership, but rejected in favor of bonds

83 6. Currently Used—Toll revenue 7. Plan to Use—Development impact fees 8. Plan to Use—Blank I.5 For what purpose(s) are the generated revenues used or planned? Please check (✓) all boxes that apply. Use of Generated Revenues Currently Use Plan to Use Facility construction 10 11 Facility operation and maintenance 15 13 Expansion of existing facility 9 8 Other roads or highways 2 4 Public transit 2 2 Debt service on bonds 14 12 General tax revenues (not specific to the facility) 1 1 Other (explanation below): 1 0 Comments related to the three respondents who selected “Other”: 1. Currently Used—Operations and investment for other [a tolling authority] facilities through consolidated bonds 2. Not Stated—Bond proceeds other roads and open space 3. Not Stated—Public partnership projects I.6 What are the current or planned toll structure and rates for your toll roads? Toll Structures Currently Use Plan to Use Point toll (e.g., expressed as $/trip) 11 6 Distance toll—fixed (e.g., expressed as $/mile) 3 3 Distance toll—variable (e.g., expressed as $/mile) 2 4 Other (explanation below): 8 Comments related to the eight respondents who selected “Other”: 1. Not Stated—[A state DOT] has no toll facilities 2. Not Stated—Per axle toll 3. Not Stated—Probably a mixture 4. Not Stated—Variable structure 5. Currently Used—Excess revenues

84 6. Not Stated—Under study at this time 7. Not Stated—Too early in development stage to determine 8. Currently Used—Time of day and class of vehicle I.7 What is your organization’s role in or use of toll road demand and revenue forecasts? Please check (✓) all boxes that apply. Responses Prepares travel demand (traffic) models/forecasts 8 Prepares or collects the data that are used for the travel demand models/forecasts 9 Conducts peer reviews or critical reviews 5 Uses the travel demand forecasts to prepare revenue forecasts or conduct a financial analysis 14 Uses the travel demand forecasts to approve or ensure funding 10 Other (explanation below): 7 Comments related to the seven respondents who selected “Other”: 1. Manage and review consulting studies 2. Revenue forecast and bond financing 3. Policy and master plans 4. Procures travel demand and finance analyst consultant services 5. Use traffic and revenue studies to establish financial feasibility of proposal project 6. Sell bond in the financial market 7. Revenue forecasts are based on historical receipts I.8 If your organization does not prepare its own toll road demand forecasts internally, who provides the forecasts for your use? Please check (✓) all boxes that apply. Responses Consultant 13 Regional MPO/COG 2 State DOT 1 Other (explanation below): 3 Comments related to the three respondents who selected “Other”: 1. A consultant will do eventually 5. Not Stated—Barrier system ~ 0.1 per mile

85 Part II. State of practice in travel demand forecasting models for toll road demand and revenue forecasts II.1 Please describe the specific application for which the model was (or is to be) used. The following information was needed to help us ensure that our survey has covered a broad range of perspectives. We also needed to follow-up with respondents for clarifications of responses. In the survey the respondent was assured that the answers to this question would be kept strictly confidential and would not be disclosed to outside parties. Therefore no responses are provided for this question. II.2 To what type of analysis was the model applied? Please check (✓) all boxes that apply. Responses Conceptual plan/feasibility plan 3 Policy study 2 Alternate analysis 6 Investment grade 6 Design forecast 3 Critical review or audit 2 Risk assessment analysis 1 Other (explanation below): 6 Comments related to the six respondents who selected “Other”: 1. Operating budget 2. State environmental analysis 3. RTP/ITP 4. Operating revenue forecast 5. Project year 2005 yearly volumes at the seven toll plazas 6. Note: RTP/ITP = regional transportation plan/intermodal transportation plan. Toll revenue and demand forecasting for existing facilities 2. Not yet determined; likely a consultant 3. N/A

86 Response #1 MINUTP 6-county [state] metro region ~ 2.5 million 2000 Response #2 TranPlan 5.3 million 2000 Response #3 Transcore proprietary and Excel Vehicles per minute Response #4 TranPlan 4,700 square miles 1.8 million 1997 calibrated annually for [a tolling authority] Response #5 Cube Voyager 3,968 square miles 3.8 million 2000 calibration/ 2005 validation Response #6 Microsoft Excel N/A N/A 2004 Response #7 TranPlan 6,288 square miles 3.2 million 2001 Response #8 TranPlan 65 square miles 9.2 million 2003 Response #9 Microsoft Excel [Two] counties bordering toll plaza 1998–2004 Response #10 Econometric models estimated in reviews and operationalized in Microsoft Excel 1,500 square miles 18 million 2003 (updated annually) Response #11 EMME/2 2,750 square miles 4,306,700 1995 Response #12 EMME/2 9,000 square kilometers (3,745 square miles) 5.5 million 1996 Response #13 EMME/2 4,200 square miles 2.5 million 2000 II.4 Please describe the characteristics of your model’s networks and zone system. No. of Traffic Zones No. of Traffic Links No. of Network Nodes Response #1 437 [No response] [No response] Response #2 3,043 55,000 20,000 Response #3 1 7 7 Response #4 2,036 27,501 12,575 Response #5 1,740 40,000 30,000 Response #6 0 1 0 Response #7 953 17,400 6,300 Response #8 3,378 35,000 25,000 II.3 Please describe the following basic characteristics of your travel demand forecasting model. Please answer all questions and/or check (✓) all boxes that apply. Software Used for the Model Area Covered by the Model Population Covered by the Model Base Year of Calibration

87 Response #12 1,700 40,000 12,000 Response #13 2,600 10,309 18,836 II.5 Please describe the types of links (tolled and not tolled) in your modelís networks. Please check (✓) all boxes that apply. Type of Links Not Tolled Tolled Expressways 9 10 Arterial roads 9 1 Collector roads 9 0 Local roads 7 0 Transit networks 3 0 Other (explanation below): 7 Comments related to the seven respondents who selected “Other”: 1. Not Stated—HOV 2. Tolled—Bridge 3. Not Stated—[State] 400 only tolled facility 4. N/A 5. Tolled—Bridge tunnel 6. Tolled—Bridges and tunnels 7. Not Tolled—Interstate (limited access) II.6 Please describe the modes that are modeled. Please check (✓) all boxes that apply. Responses Single-occupancy passenger vehicle (SOV) 9 High-occupancy passenger vehicle (HOV) 7 Vehicles in commercial use (e.g., repair vans, taxis, courier trucks, etc.) 4 Trucks (light and heavy) 8 Emergency/military vehicles 0 Buses 5 Passenger rail 3 Response #9 No formal model No formal model No formal model Response #10 [No response] 8 toll plazas [No response] Response #11 986 61,000 one-way 15,000

88 2. All vehicle types listed above are forecasted, but distinct models are used to project revenue-paying autos, buses, light trucks, and heavy trucks. 3. Passenger cars with no specified vehicle occupancy II.7 Please describe the time periods modeled. Please check (✓) all boxes that apply. Responses Weekday AM peak hour 9 Weekday PM peak hour 8 Weekday other peak 2 Weekday off-peak 8 Weekday 24 hour 5 Weekend 2 Other (explanation below): 7 Comments related to the seven respondents who selected “Other”: 1. Weekday night period 2. Peak season weekday 3. Aggregate monthly 4. Entire 365 day year 5. All time periods broken out for each vehicle-type forecast; however, distinct models by time period are factored by relevant activity data to arrive at time-of-day forecasts. 6. Midday (9 a.m.–3 p.m.) and nighttime (6 p.m.–6 a.m.) 7. Nighttime (NT) II.8 Please indicate how the peak hour is derived. Please check (✓) one box only. Responses Modeled directly in demand model (no factors) 3 Factors or percentages applied to trip tables before assignment 4 Other (explanation below): 5 Non-response 1 Other (explanation below): 3 Comments related to the three respondents who selected “Other”: 1. Airport

89 4. Provided by MPO 5. Percentage applied to peak period to get peak hour II.9 Please describe the model structure. Please check (✓) one box only. Responses Traditional four-step (generation, distribution, modal split, assignment) 4 Trip assignment only (no demand modeling, or demand is forecast externally to the model) 3* Activity-based modeling 1 Other (explanation below): 5 Non-response 0 *Note: One respondent to this option (trip assignment only) commented that the “MPO provides trip tables obtained from the regional model.” Comments related to the five respondents who selected “Other”: 1. Traditional four-step plus toll diversion 2. Projection 3. Existing volumes, growth rate, estimated increase/decrease due to specific development or detour 4. Econometric model for traffic projection by vehicle type and facility type 5. Non-traditional four-step with toll trips as mode choice II.10 Does the model have feedback loops? Please check (✓) one box only. Responses Yes—assignment impedances are fed back to distribution and/or modal split for cycles 6* Yes—other 0 No feedback loops 7 Non-response 0 *Note: Two respondents that selected the first option commented as follows: (1) “1 cycle for mode share only,” and (2) “Yes—other → Mode Choice.” Comments related to the five respondents who selected “Other”: 1. Peak hour not modeled 2. Factors applied to daily assignment 3. N/A

90 Shopping 7 Leisure/recreation 6 Personal business (e.g., medical appointment) 5 Serve passenger (pick up/drop off) 2 Commercial vehicles (i.e., trucks, etc.) 2 Other (explanation below): 8 Comments related to the eight respondents who selected “Other”: 1. Home-base other, non-home-base, air [passengers] 2. Home-base other/non-home-base other 3. Major tourist centers are special generators 4. N/A 5. Work/business-related and non-work/other 6. None 7. All trip purposes are modeled by vehicle type (auto, bus, light truck, heavy truck) for each crossing 8. Non-home-based non-work (NHBNW), home-base other (HBO) II.12 Please describe the formulation of the modal split model (methods for mode choice modeling). Please check (✓) all boxes that apply. Responses Logit or similar 6 Diversion curve 0 Factors 4 No modal split model 3* Other (explanation below): 3 *Note: One respondent to the fourth option (no modal split model) commented that “Assignment based on vehicle trip tables.” Comments related to the 3 respondents who selected “Other”: II.11 Please describe the trip purposes modeled. Please check (✓) all boxes that apply. Responses Work (commute) 7 Work-related 7 Out-of-town business 2 School/education 6

91 Responses Equilibrium assignment 8 All-or-nothing 1 Capacity restraint 1 Other 2 Non-response 1 Comments related to the two respondents who selected “Other”: 1. N/A 2. Capacity and time-value of motorists to select nearby toll bridge or further free bridge II.14 Please describe the methods used for time choice modeling. Please check (✓) all boxes that apply. Responses Peak spreading model 1 Time choice model 1 Arrival/departure time choice model 0 Factors from surveys, counts, or other sources 6 No time-of-day choice models 4 Other (explanation below): 2 Comments related to the two respondents who selected “Other”: 1. N/A 2. Mode choice is run by the MPO II.15 Please describe the tolling costs that are modeled. Value of Time* Willingness to Pay* Other* Mode 4 0 0 1. N/A 2. Trip tables provided by MPO 3. Revealed preference (origin–destination) survey II.13 Please describe the formulation of the trip assignment model (methods for route choice modeling). Please check (✓) one box only.

92 2. Vehicle Class—By vehicle class. Toll and fuel prices are controlled for in the econometric model. 3. Purpose—Travel time, travel cost, and income modeled as mode choice utility equations classified by trip purpose and time of day. II.16 Are variable tolls modeled? Please check (✓) one box only. Responses Yes 3 No 10 Non-response 0 II.17 How are tolls taken into account in trip assignment (route choice)? Please check (✓) all boxes that apply. Responses Generalized cost in volume-delay function 4 Diversion curve 5 Other (explanation below): 3 Comments related to the three respondents who selected “Other”: 1. N/A 2. Generally following existing automobile choice from existing data 3. Toll elasticities are estimated from observed response to toll changes and variable pricing. II.18 Was this model calibrated specifically for this analysis? Or was it adapted from another model? Please check (✓) all boxes that apply. Responses New model was calibrated specifically for this toll demand and revenue forecast 2 Vehicle class 4 2 2 Purpose 4 2 1 *Note: One respondent who did not select any option commented “N/A.” Comments related to the three respondents who selected “Other”: 1. Vehicle Class—Automobiles have choice to use lightweight bridge in some cases. Trucks generally do not have a nearby alternative.

93 1. NTTM model is supported by information provided by the MPO II.19 Are there any special or innovative features related to this application, not otherwise described above? Response #1 Incorporated toll diversion curves into previously calibrated model Response #2 The [state] toll facilities model is used to address queuing and delay at plazas Response #3 Toll diversion model Response #4 Trip reduction/consolidation factors due to tolling existing free bridge crossing Response #5 The model is updated periodically to account for the behavioral changes in the [state] metropolitan area. Response #6 This is the sixth annual projection completed. After the end of the year, actual toll volume is compared with projected toll volume revenue for the year, and growth rates are adjusted accordingly. Response #7 The model allows scenario development/sensitivity testing of economic variables and assumptions regarding electronic payment utilization. A separate value pricing model has been developed that allows for estimation of traffic/revenue impacts of pricing changes by vehicle class, hour, method of payment, and crossing. Response #8 A major shortcoming of a planning-level regional model is that traffic operations on each link operate independently of every other link. That is, there is a certain level of delay associated with a link that results from the traffic volume and link capacity only. The reality of external influences such as traffic from merging roadways or downstream blockages are not reflected in the regional model since queuing impacts from these situations are an important component of travel delay along the expressway during peak hours, it was necessary to modify link capacities in those areas where queue buildup is significant. Response #9 Auto trips split into toll and non-toll components by trip purpose and time of day in the mode choice step Note: Nine of 13 respondents to Part II answered this question. II.20 What data were used to calibrate or validate your model? Please check (✓) all boxes that apply. Model was updated/enhanced from an existing model that was calibrated for other purposes (e.g., LRTP, TIPs, etc.) 6* Existing model was used as is without special adaptations 1 Don't know 0 Other (explanation below): 1 *Note: One agency that selected the third option commented “[Toll Authority] revalidates annually.” Comments related to the one respondent who selected “Other”: Model was updated/enhanced from an existing model that was calibrated for previous toll demand and revenue forecasting 4

94 Comments related to the five respondents who selected “Other”: 1. Calibration—Electronic toll collection perception 2. Calibration—Characteristic of toll users 3. Validation—Toll impedance 4. Validation—Electronic toll collection perception 5. Validation—Characteristic of toll users II.21 What tests were used to validate the base year model assignment results? Please check (✓) all boxes that apply. Responses Comparison of ratios of simulated and observed volumes at screenlines, cutlines, cordons, etc. 10 Statistical tests (R sq., RSME, GEH, etc.) 7 Comparison of simulated and observed travel times or speeds on links, facilities, corridors, etc. 8 No tests/don’t know 2 Other (explanation below): 3 Comments related to the three respondents who selected “Other”: 1. K factors adjusted to reflect origin–destination 2. Compared actual vs. projected volumes from previous study 3. Sensitivity analysis II.22 Please indicate how your model and/or toll road forecasts have been verified. Please check (✓) all boxes that apply. Responses None/not done 5* Origin–destination survey 5 7 Activity- or tour-based survey 2 1 Stated preference survey 3 3 Traffic counts 8 10 Speed/travel time surveys 3 9 Land use inputs 2 6 Network characteristics 3 6 Other (explanation below): 2 3 Data Calibration Validation

95 1. Validated against existing conditions 2. Observed traffic II.23 Were changes, corrections, or improvements recommended for the model or the forecasts as a result of the verification? Responses Yes (explanation below) 5 No 3 Non-response 5 Comments related to the five respondents who selected “Yes”: 1. [No response] 2. The model is updated periodically 3. Econometric models were respecified when realistic out-year forecast growth rates were not achieved by vehicle type. Quarterly forecasts are now being developed to refine annual forecast errors and improve on variance reporting throughout the year. 4. [No response] 5. Model refinements in validation II.24 Were these recommendations implemented? Responses Yes—All 4 Yes—Some (explanation below) 1 No—Not implemented in current model forecasts 0 Planned for implementation in future models and forecasts 0 Judgment/reality check 8* Statistical verification 2 Risk assessment 2 Sensitivity tests of key parameters or inputs 7* Use of ranges of forecasts 1 Independent audit/critical review 2 Other (explanation below): 2 *Note: One respondent to the first, (none), second (judgment), and fifth (sensitivity tests) commented that “Tolling not in place yet.” Comments related to the two respondents who selected “Other”:

96 No (explain) 0 Non-response 5 *Note: One respondent to the first option commented that “Bonds have been rated on 3 occasions.” II.26 Were network micro-simulation models used in the development of demand and revenue forecasts (e.g., for a HOT lane)? Responses Yes 0 No 8 Non-response 5 II.27 What other type(s) of models were used for this application? Please check (✓) all boxes that apply. Responses Land use/economic forecasting model (software used listed below) 2 Traffic operations model (software used listed below) 1 Other (explanation below): 3 None 1 Comments related to the six respondents who selected “Land Use/Economic Forecasting Model” or “Traffic Operations Model” or “Other”: 1. Land use/economic forecasting model—Proprietary 2. Traffic operations model—HCM for the environmental studies 3. Land use/economic forecasting model—DRAM/EMPAL 4. Other—Independent economic review Non-response 8 Comments related to the one respondent who selected “Yes—Some”: 1. Additional data are needed for high-speed electronic toll collection II.25 Have the results of your forecasts been accepted by the intended audience/decision makers? Responses Yes, unconditionally 7* Yes, with qualification of conditions 1

97 conditions Long-term (>10 years) forecasts vs. actual conditions 0 0 2 Corresponding ramp-up percentage 50% — — Corresponding medium-term percentage — 24% — Corresponding long-term percentage — — — III.2 What impact did the differences (if any) identified in Question III.1 have on the forecasts or on the use of the forecasts? Please check (✓) all boxes that apply. Responses Model was recalibrated/model networks were reconfigured 2 Demand forecasts were revised 2 Revenue forecasts were revised 2 Financial schedule was revised 1 Performance indicators changed/new indicators implemented 0 Tolling structures or rates were changed 0 Project costs changed 0 Staging or timing of project was revised 1 Project postponed or canceled 1 Policies revised/new policies adopted 0 No impact (forecasts accepted and used as is) 7* Other (explanation below): 2 *Note: One respondent to the eleventh option (no impact) commented “Facility not tolled yet.” Comments related to the two respondents who selected “Other”: 5. Other—National/regional economic forecast variables from Global Insight and Economy.com 6. Other—A logit modeling software Part III. Experience with toll road demand and revenue forecasts III.1 Please describe how your toll road demand forecasts have compared with actual demand conditions or with the expected performance of your facility. Traffic Forecasts Overstated Traffic Forecasts Understated Traffic Forecasts within 5% Ramp-up forecasts vs. actual conditions 1 0 7 Medium-term (5–10 years) forecasts vs. actual 0 1 4

98 Public and political inputs regarding land use and network assumptions 2 Environmental or economic development considerations 2 Other (explanation below): 1 Comments related to the one respondent who selected “Other”: 1. Economic climate Modeling: Responses Model structure 4 Process or expanding modeled time periods to annual forecasts 4 Transparency/opacity in the modeling and forecasting processes 1 Calibration process, coverage, and precision 2 Some modes were not modeled (e.g., trucks) or were not modeled well 0 “Control” over how the model outputs were used, analyzed, or interpreted 2 Other (explanation below): 1 Comments related to the one respondent who selected “Other”: 1. Validity of models for financing purposes Operations: Responses Staging or proposed facility (or other facilities) 2 1. Annual updates and peer reviews 2. Anticipated revenues are used to determine upcoming revenues. If revenues are overstated, some projects may be postponed. III.3 What factors influenced the performance of the forecasts identified in Question III.1? Please check (✓) all boxes that apply. Inputs: Responses Assumptions regarding land use or future base network configurations on parallel/competing routes or modes 5 Availability, appropriateness, or sufficiency of data, models, or analytical capabilities 4 Values used for value of time/willingness to pay and/or other monetary values 6

99 bottom of the inputs table) Note: Two of 13 respondents to Part III answered this question. III.4 What recommendations do you have to improve the usability, accuracy, reliability, and credibility of travel demand models and forecasts for estimating toll revenues and for financing? Please check (✓) all boxes that apply. Responses Improve transparency in the modeling and forecasting processes (so that decision makers are better informed) 4 Improve methods for travel demand forecasting modeling 8 Collect better or more data as the basis for model calibration or to monitor conditions 8 Provide better training for modeling and planning staff 5 Find better ways to tie modeling process to organizational/facility business or financing plan 3 Conduct more risk assessments in forecasts 7 Conduct more critical reviews and audits 4 Other (explanation below): 1 Comments related to the one respondent who selected “Other”: 1. More direct relationship to economic factors that determine travel demand III.5 What recommendations from previous forecasting applications did this model (for this specific application) already incorporate? Please check (✓) all boxes that apply. Actual operations and system reliability (e.g., congestion levels, operating speeds, frequency of incidents, etc.) 4 Impact of tolling technology on actual traffic volumes (e.g., unreadable license plates) 4 Violation rate 3 Changes in policy, mandate, legislation, ownership, political environment, etc. 1 Other 0 Additional Comments: Response #1 Consideration of the factors above were among the reasons the [state toll road] forecast has been successful [factors checked included inputs table (first three responses reading from top to bottom), modeling table (the first four and the six responses), operations table (first four responses)]. Response #2 Not a standard link–node model (only check the 1 and 3 responses reading from top to

100 III.6 Which of these recommendations do you plan to implement in your models or in your next application? Please check (✓) all boxes that apply. Responses Improve transparency in the modeling and forecasting processes (so that decision makers were better informed) 1 Improve methods for travel demand forecasting modeling 5 Collect better or more data as the basis for model calibration or to monitor conditions 7 Provide better training for modeling and planning staff 1 Find better ways to tie modeling process to organizational/facility business or financing plan 2 Conduct more risk assessments in forecasts 4 Conduct more critical reviews and audits 1 Other 0 III.7 What factors would prevent you from implementing these improvements? Please check (✓) all boxes that apply Responses Lack of financial resources 5 Lack of staff 1 Lack of time 4 Don’t know how to go about it 0 Not in our mandate 1 Other priorities 5 Improve methods for travel demand forecasting modeling 4 Collect better or more data as the basis for model calibration or to monitor conditions 6 Provide better training for modeling and planning staff 2 Find better ways to tie modeling process to organizational/facility business or financing plan 1 Conduct more risk assessments in forecasts 1 Conduct more critical reviews and audits 2 Other (explanation below): 2 Comments related to the two respondents who selected “Other”: 1. Simplicity 2. More direct relationship to economic factors that determine travel demand Responses Improve transparency in the modeling and forecasting processes (so that decision makers were better informed) 3

101 Response #1 The adaptation of the MPO models for use by this agency has been refined over a 10-year period. The forecasts results are review monthly against actual reviews. This application has been brought before the rating agencies on numerous occasions and forecast review have not been a bond rating issue as the forecasts are consistently 2%–5% lower than actual. The agency recently received a rating agency upgrade on the strength of the last bond presentation and the appropriately conservative nature of the revenue forecasts. The process of forecasting revenues benefits from a built-in peer review process and rigorous annual updates used for the traffic engineers annual revenue forecasts. These forecasts are used to set the agencies annual operating budget. Every year, a specific aspect of the model is focused on. Examples include sub-area land use updates, new data on high-speed electronic toll collection performance, or updates to remain consistent with the regional transportation models. Response #2 [A tolling authority] does not employ a computer-based travel demand and forecasting model. Rather, we perform manual projections and analysis to support our current operations and possible changes to toll structure. Response #3 Most of this survey does not apply directly to [bond rating agency], and we have experience working with forecasts generated by too many different models to address the problems we have had with each of them individually, but I think that most of the recommendations in this survey would be helpful for most if not all of the forecasts we have worked with, especially for start-ups. Response #4 Part II asks for details on “state-of-the-practice” travel demand forecasting models. Part III asks for ramp-up, medium and long-term model results of the same model described in Part II. These parts are in conflict. Traffic and revenue forecasts leading to project decisions are prepared years before those projects are chosen, designed, and constructed. The required lead time is easily 10 years. So to be able to answer Part III, the survey asks us to pick a model application that is at least 10 years and most likely +20 years old. Models of such lineage cannot be considered as “state-of-the-practice” by any stretch of the imagination. We have chosen to respond to Part II with what we consider to be our most advanced model development (i.e., our state-of-the-practice) achievements. The traffic and revenue forecasts produced in this manner are for projects that have no traffic history simply because they are still in production. Therefore, we are unable to provide corresponding “state-of-the-practice” responses for Part III. Note: Four of 13 respondents to Part III answered this question. None 2 Other 0 III.8 Are there other models that were used for other applications that you would like to describe? Responses Yes 0 No 10 Non-response 3 Are there any other comments that you would like to make, either on topics that have not been addressed earlier or to amplify or clarify what you have already said?

Response #9 [State] toll roads were built in the 1950s, 60s, and 70s. Of the 10 constructed, tolls remain only on 2. There have been discussions involving the future use of tolls on existing facilities and also on project tolls; however, no detailed analysis has been undertaken at this time. Response #10 [State] is planning on owning/operating a toll facility with [a state DOT] when they either get funds to complete their portion or when state law allows them to own/operate a toll facility. Other comments were made by respondents who only completed Part I or Part II of the survey. These are documented here: Response #5 [State DOT] is presently in the process of selecting consultants to perform toll feasibility studies for three locations. We do not now have any toll facilities owned/operated by the state. Response #6 We are new. We would like a copy of the results when complete if possible. Response #7 [A state DOT] has only three toll facilities: [three bridges]. Our statewide travel demand model has not recently been used to forecast traffic over these crossings. Traffic forecasts for future studies at proposed new border crossings will probably be the subject of special contracts, and will probably not be sensitive to the impact of toll rates, since there are no untolled alternatives to the crossings: Our statewide model contains no provision for testing toll alternatives or making toll-road forecasts, other than to change the impedance on links proposed as toll routes. It has never been used for this purpose. No toll-road projects are under study in Michigan, and none are foreseen. Response #8 No consultant performing toll forecasting at this stage of the project 102