BLUE RIBBON PANEL

1
Expert Forum on Road Pricing and Travel Demand
Modeling
Modeling Pricing in the Planning Process
Ram M. Pendyala Department of Civil and
Environmental Engineering University of South
Florida, Tampa U.S. Department of
Transportation Alexandria, VA November 14-15,
2005
2
Outline

• Introduction and Motivation
• Role of Travel Demand Modeling
• Variety of Pricing Mechanisms
• Road Pricing Projects U.S. and Abroad
• Pricing and Network Dynamics
• Experiences with Toll Road Forecasting
• Sources of Errors in Forecasts
• Four/Five-Step Travel Demand Models

3
Outline (continued)

• Key Behavioral Processes Underlying Response to
  Pricing Policies
• Advances in Travel Demand Modeling Methods and
  Paradigms
• Conclusions and Future Directions

4
Introduction and Motivation

• Pricing and innovative toll strategies
• Drivers pay marginal cost of travel congestion
  and externalities
• Travel demand management strategy
• Reduce auto travel mode destination shifts
• Suppress auto travel eliminate or combine trips
• Reduce peak period congestion temporal shifts
• Revenue generation
• Invest in transport infrastructure improvements
• Pay off debt
• Desire for high volumes of paying users
• Conflicting objectives?

5
Planning Methods for Pricing Strategies

• Sketch planning techniques
• Elasticity methods
• Peer city comparisons
• Similar facility comparisons
• Stated preference research
• Estimates derived from stated preference data
• Travel demand modeling systems
• Variations of four-step travel demand modeling
  methods
• Forecast patronage, traffic impacts, and revenue
  stream into future

6
Pricing-Strategy Related Impacts

• Traffic and travel demand impacts
• VMT, VHT, travel time, delay, traffic volumes
• Accessibility impacts
• Revenue generation perspective
• Patronage or volume of demand by time of day
• Market penetration by payment type/technology
• Short- and long-run demand elasticities
• Social equity and environmental justice
• Mobility, accessibility, and economic impacts by
  market segment (income, car ownership, gender,
  age, etc.)

7
Variety of Pricing Mechanisms

• Public transport pricing systems
• Parking pricing
• Standard (flat) tolls
• Shadow tolls
• Area-Based/Distance-Based Congestion Charging
• Variable/Dynamic/Value Pricing/Tolls
  Facility-Based
• HOT Lanes/FAIR Lanes
• Credit-based congestion pricing

8
Road Pricing Projects U.S. and Abroad

• FHWAs five types of value-pricing projects
• A. Pricing on existing roads
• B. Pricing on new lanes
• C. Pricing on toll roads
• D. Pricing of parking and vehicle use
• E. Region-wide studies/initiatives
• Several operational and others under study
• Considerable international experience
• Singapore 25 years of experience
• Central London 2-3 years of experience

9
Pricing and Network Dynamics

• Optimizing traffic networks using pricing
  mechanisms
• Minimal-revenue congestion pricing to induce
  system optimal performance
• Dynamic traffic network simulation
• Variety of electronic toll/pricing technologies
• Mix of users changes over time
• Modeling impacts of variable pricing requires
  explicit recognition of network dynamics

10
Pricing Project Experiences

• Several projects described in paper
• SR 91 express lanes in California
• San Diego I-15 congestion pricing project
• Lee County (Florida) variable pricing project
• Singapore congestion pricing implementation
• Central London congestion charging scheme
• All projects report various degrees of success
• Decrease in traffic congestion, particularly in
  peak periods
• Substantial patronage/usage of toll facilities

11
Toll Road Forecasting Experience

• Toll road forecasts with traditional travel
  demand model systems
• Minor variations to incorporate sensitivity to
  pricing
• Analysis of toll road forecast accuracy
• Toll road forecasts overestimated traffic by
  20-30
• Review of 87 toll road projects Average ratio of
  actual/forecast patronage is 0.76
• Suggest presence of significant systematic
  optimism bias
• Previous experience with toll facilities helps
  improve accuracy of forecasts

12
Sources of Errors in Forecasts

• Errors in socio-economic and land use forecasts
  that serve as inputs to model system
• Errors in input assumptions including model
  coefficients, costs, rates, value of travel time
• Errors in coding networks and node/link
  attributes by time-of-day
• Errors in truck travel forecasts
• Errors in estimate of ramp-up period
• Errors in behavioral paradigms underlying travel
  demand forecasts

13
Induced/Suppressed Travel

• In response to pricing
• Trips may be eliminated due to additional cost
• New trips may be induced due to improved
  level-of-service
• Traditional models unable to account for impacts
  of accessibility on trip generation (activity
  participation)

14
Trip Chaining and Tour Formation

• In response to pricing
• Trips may be combined/linked into chains/tours
• Additional cost may induce desire for efficiency
• Shifts in trip timing may lead to trip chain
  formation
• Need to recognize inter-dependencies among trips
  in a chain (e.g., mode, destination)

15
Time-Space Geography

• Behavioral response to pricing strategies
  influenced by
• Spatio-temporal flexibility and constraints
• Defining time-space prisms
• Time allocation and time use behavior (activity
  episode duration)
• Scheduling/timing of activities and trips
• Time of day modeling along the continuous time
  axis

16
Agent-Based Interactions and Inter-dependencies

• Traveler response to pricing strategies dependent
  on host of interactions
• Interactions among household members activity
  allocation and joint activity engagement behavior
• Activity scheduling and re-scheduling behavior
• Inter-dependencies among activities and trips in
  a complete activity-travel pattern
• History dependency and inter-temporal
  relationships
• In-home out-of-home activity substitution and
  complementarity

17
Secondary/Tertiary Impacts

• Primary impact on specific trip(s) subjected to
  pricing strategy
• Interactions/inter-dependencies result in host of
  secondary/tertiary impacts
• Complete activity-travel pattern subject to
  change as trips are
• rescheduled and chained
• shifted in time, mode, destination, route
• Impacts on other household members

18
Microsimulation Approaches

• Simulation of complete activity-travel patterns
  for each individual in population
• Modeling at the level of the individual
  decision-maker
• Represent behavioral decision-making processes
• Capture differences (taste-variation) across
  individuals
• Synthesize and evolve population over time
• Reflect population dynamics
• Ramp-up period represents evolutionary period of
  learning and adaptation

19
Dynamic Traffic Assignment

• Pricing policies increasingly variable/ dynamic
  in nature
• Travel times, costs, paths, and speed-flow
  patterns constantly updated
• Dynamic traffic assignment algorithms to reflect
  network dynamics
• Integrate with activity-based models
• Appropriate feedback loops network impacts on
  activity patterns

20
Integrated Urban Systems and Activity-Travel
Modeling

• Host of medium and longer term choices
  potentially impacted by pricing policies
• Residential and work location choice
• Vehicle ownership choice
• Business location choice
• Changes in property values and land accessibility
• Evolution of urban system
• Feedback between activity-travel demand model and
  land use simulation model

21
Heterogeneity in Population Attributes

• Heterogeneity in population attributes
• Attitudes and perceptions towards pricing
  strategies
• Preferences for and values attributed to
  alternative behavioral responses
• Values of travel time savings and travel time
  reliability
• Learning and adaptation strategies
• Recent advances in econometric model formulation
  and estimation
• Presence of heterogeneity in value of travel time
  savings proven

22
Role of Attitudes and Perceptions

• Attitudes and perceptions shape behavior (and
  vice-versa)
• Nature and magnitude of response to pricing
  policy
• Adaptation strategies adopted
• New activity-travel pattern considered
  acceptable or satisfactory or optimal
• Adoption of electronic toll collection
  technologies
• Habitual vs. occasional use of tolled facility
• Help inform model framework, behavioral paradigm,
  and model specification

23
Towards a New Generation of Modeling Approaches

• Tour-based and activity-based microsimulation
  model systems
• Advanced econometric model estimation methods
• Reflect behavioral decision-making processes
• Cause-and-effect relationships
• Integrated modeling of land use activity/travel
  demand traffic network continuum with feedback
• Long-term to short-term choices
• Not necessarily unique to pricing policies many
  other emerging behavioral, policy, technology,
  and environmental issues

24
Pricing Considerations

• Unique nature of pricing schemes that amplify
  issues with models
• Direct cost/monetary implications
• Direct travel time/reliability implications
• Direct infrastructure finance implications
• Absence of incorporation of monetary constraints
  (expenditures vis-à-vis income)
• Some decrease in VMT growth, but generally little
  (short-term) impact of fuel price rise

25
Pricing Considerations (continued)

• What should toll reflect/accomplish?
• Value of travel time savings
• Value of travel time reliability
• Facility construction/maintenance costs
• Congestion/externality costs (full cost pricing)
• Network-wide ripple effects
• Shifts to facility due to improved LOS
• Shifts away from facility due to added cost
• Shifts to improved toll-free facilities

26
Hierarchy of Behavioral Response?

• Modify attribute of least impact first?
• Route shift
• Temporal shift
• Trip chaining shifts
• Destination shifts
• Mode shifts
• Activity (re)allocation
• Activity participation (elimination/addition)
• Auto ownership
• Workplace/residential location
• Implications for behavioral modeling

27
Key Opportunities

• Widespread interest in implementation of
  innovative pricing schemes/technology systems
• Toll road forecasts coming under intense scrutiny
• Determine contribution of various sources of
  error
• Input data/assumptions/variable forecasts
• Model specifications/parameters/variables
• Behavioral paradigm/framework
• Heterogeneity in traveler perceptions and values

28
Key Opportunities

• Several real-world projects offering data on
  observed behavior
• Conduct longitudinal surveys of behavior in
  conjunction with ongoing projects
• Test and validate advanced travel demand modeling
  methods
• Controlled studies involving comparisons of
  forecasts offered by different modeling methods
• Special experiments to understand behavioral
  adaptation, heterogeneity, and attitudes/perceptio
  ns