Evaluation of TimeofDay Fare Changes for Washington State Ferries

1
Evaluation of Time-of-Day Fare Changes for
Washington State Ferries
Prepared for TRB Transportation Planning
Applications Conference
2
Presentation Outline

• Project background
• Project objectives
• Study approach
• Results
• Conclusions

3
Washington State Ferries Background

• Formed in 1951, is the largest ferry transit
  system in the U.S.
• Serves about 23 million passengers and vehicle
  trips per year
• Operates 10 ferry routes and runs nearly 500
  sailings per day
• Provides service to eight Washington State
  counties and the Province of British Columbia
• Operates and maintains 20 terminals
• Provides priority loading for bicycles, vanpools,
  and carpools

4
Washington State Ferries Long Range Plan

• Plan is based on a 2007 legislation and needed
  to
• Develop strategies to minimize capital and
  operational costs
• Implement adaptive management practices to
  improve service quality and keep costs at lowest
  possible level
• Re-establish vehicle LOS standards
• Plan developed jointly by WSF staff States
  Joint Legislature Transportation Committee with
  input from Washington State Transportation
  Committee
• Plan has been submitted to legislature for review
  and is being finalized

5
Washington State Ferries Travel Demand Model

• Initially developed in early 1990s and updated
  using 1993, 1999 2006 ferry travel survey data
• Focuses on PM peak ferry travel
• Covers 12-County service area
• Uses incremental methods (as in Sound Transit
  model)
• Relies on observed ferry travel patterns
• Interfaces with PSRC model transportation and
  land use data from other jurisdictions

6
Purpose of Our Work

• Evaluate the effects of fare policy changes on
  ferry traffic
• Overall fare increases
• Changing car/walk-on fare differentials
• Time-of-day fare differentials
• To do this, needed to estimate fare elasticities
  for
• Using the ferry
• Ferry submode
• Time-of-day shifts

7
Study Approach

• Designed stated preference survey to collect
  information on likely responses to fare and other
  service changes
• Four attributes fare, waiting time, time of
  earlier sailing, time of later sailing
• Five choice alternatives
• Drive-on ferry at current sailing time
• Drive-on ferry at earlier sailing time
• Drive-on ferry at later sailing time
• Walk-on ferry at current sailing time
• Use other non-ferry alternative
• Survey was administered to 840 current drive-on
  customers
• Data from survey were used to estimate discrete
  choice models
• Aggregate models to determine appropriate
  segmentations and specifications
• Individual-level models estimated using mixed
  logit and hierarchical Bayes

8
Some Notes on Choice Modeling Approach

• Market research firm that collected data
  conducted initial choice modeling
• Used hierarchical Bayes estimation
• Provides individual-level utilities
• Used different type of specification
• Attributes-only (no systematic sources of
  heterogeneity)
• Over-specified model
• Represented fare with eight discrete levels
• Models showed much higher fare elasticity than
  seemed reasonable
• The WSF team developed refined choice models
• Specified continuous fare utility functions
• Allowed non-linear income effect on fare
  sensitivity
• Reduced specifications to ones that could be
  identified
• Segmented models to allow more consistent priors
• Estimated models with both mixed logit and
  hierarchical Bayes
• Produced posterior (individual-level) utility
  functions

9
Refined Choice Model Results

• Choice models indicate that drive-on customers
  are willing on average to shift departure times
  by 30 minutes for a 3 fare savings
• Discretionary trips have more flexibility in
  departure times
• Some differences in flexibility among routes
• Spreadsheet-based simulation used to calculate
  route-group and segment elasticities
• Overall drive-on fare elasticity estimated using
  stated preference data is closely comparable to
  observed historical fare elasticity
• Stated preference fare elasticity -0.30
• Historical fare elasticity -0.32
• Average elasticity of fare sensitivity to income
  is -0.13

10
Elasticity Estimates by Segment
Elasticity of Peak Drive-on Volume to Off-peak
Fares (20 off-peak fare decrease)

• Overall elasticities to time-of-day fare changes
  somewhat higher than expected
• Relatively high current fares
• Higher income customers have more time flexibility

11
Effects of Differential Time-of-Day Fares
Time-of-Day Fare Sensitivities

• Modest peak period differentials cause
  significant enough shifts to relieve peak hour
  capacity issues

12
Higher Fares with Increased Walk/Drive
Differences

• Significant mode shifts can be induced by pricing
  changes

13
Higher Fares with Increased Walk/Drive
Differences

• Revenue increases with drive-on fare increases up
  to 50

14
The Current Washington State Ferries Plan
The Plan proposes use of a reservation system to
manage demand for the limited peak-period
drive-on capacity. It proposes encouraging
walk-on use by increasing passenger fares at half
the rate of vehicle fares. The Plan also
discusses other pricing strategies including
time-of-day-based congestion pricing for
possible future consideration after first
implementing the reservation system. The Plan
notes that The pricing strategy with the
greatest potential to shift travel behavior is
congestion pricing. If reservations alone are not
sufficient to shift demand then it may be
necessary to evaluate a reservations plus
variable congestion pricing approach.
The Washington State Ferries Draft Long-Range
Plan is available online at www.wsdot.wa.gov/ferr
ies/planning/ESHB2358. Appendix D of the Plan
contains details of the time-of-day elasticity
estimation process and results.
15
Conclusions

• Price elasticity estimates developed using stated
  preference survey data and properly specified
  discrete choice models are comparable to those
  estimating using historical data
• Stated preference surveys can in addition be used
  to estimate time-of-departure and submode
  shifting elasticities
• In the Washington State Ferries markets, the
  elasticity of departure time with respect to fare
  is reasonably high. This means that variable
  pricing can be a very effective means of shifting
  customers away from peak departure times
• While the Washington State Ferries Draft
  Long-Range Plan does not currently propose use of
  variable pricing, it does say that variable
  pricing may be considered in the future if other
  demand management techniques are not sufficient
  to achieve system objectives.