Edward Robson
Research Centre for Integrated Transport Innovation (rCITI)
University of New South Wales
Supervisor: Dr Upali Vandebona
Consumer Benefit Model forDeveloping Public Transport Systems
• What happens when we want to improve the transport network?
• Is there a more efficient way to do this?
2/14
Introduction
Set of options
Transport modelling
Economic analysis
Development of final option
• Aim: to develop a model to integrate the economic evaluation of a transport network alteration with a transport demand model
• Purpose: to perform rapid appraisals of consumer benefit for transport network design
• Case study: evaluation of a metro network proposal for Sydney
3/14
Details of the study
• Who is affected by the transport network?• Consumers• Suppliers• Community
• Consumer surplus:
The transport market
4/14
P (price)
D (demand)
D (P)
P0
CS
0
Pm
P (price)
D (demand)
D (P)
P0
CS
0
Pm
P1
ΔCS
• What do consumers prefer?• Lower fares• Shorter travel time – in-vehicle time, waiting time,
access time• Better comfort etc.
• Price of transport generalised cost of transport (GC):
• To calculate consumer surplus at an aggregate level, we need:
• A demand model• Generalised costs across the network
5/14
Modelling consumer benefit
• Demand for each mode in each origin-destination pair can be predicted with the multinomial logit model:
• Consumer surplus is calculated using the logsum:
• Linear approximation (rule-of-a-half) is most common
6/14
Modelling transport demand
• For each origin-destination pair in the network:1. Measure generalised costs for each mode before the
network change
2. Measure generalised costs for each mode after the network change
3. Calibrate scale factor μ
4. Measure change in consumer surplus using the logsum
• Sum the changes in consumer surplus for every origin-destination pair
• Key assumption:• Origins and destinations of trips do not change
7/14
The final model
8
413 O-D pairs48,966 people
• Aim: to calculate the change in consumer surplus, per AM commute period, from introducing a metro network to Sydney
• Maroubra to Drummoyne; Strathfield to St Leonards
• 1km study area radius surrounding each station
• Analysis:• Base case analysis using logsum
• Sensitivity tests of metro network form; metro network service; generalised cost parameters; analytical procedure
• Logsum results compared with rule-of-a-half
9/14
Case study
10/14
Methodology
Generalised costs of existing network
Generalised costs of network with
metro
• Public transport: measured using Google Maps
• Other modes: back-calculated using Journey to Work data
• Public transport: metro, as modelled on Hong Kong network
• Other modes: kept identical to previous
• Scale factor calibration and consumer surplus calculation implemented with Microsoft Excel macro
LogsumΔCS#
Rule-of-a-half ΔCS#
Increase in public
transport usage
Base case* $63,997 $63,401 30.60%
Increase VOT by 100%
$127,997 $126,803 30.60%
Set μ = 0.1 $56,443 $56,414 9.00%
Set μ = 1 $77,636 $72,446 57.30%
11/14
Results
* Average μ in base case was 0.336
# Per morning commute period
• Rule-of-a-half results were around 1% lower than logsum results
• Sensitivity tests of network parameters behaved predictably
• Scale factor has a large influence – logsum results appear to converge towards rule-of-a-half results as the scale factor approaches 0+
• A number of simplifying assumptions were required to calculate consumer surplus with the available data, e.g. waiting times
• Accuracy could be improved by using more detailed models for generalised cost parameters
12/14
Discussion
• Can measure the consumer surplus of any changes to the network, as long as they are reflected in generalised costs, e.g.
• Fares• Travel time• Frequency• New routes
13/14
Uses of the model
• Final model rapidly calculates consumer surplus from a transport network change, using an integrated transport model
• Only requires a small amount of data and accounts for changes in consumer surplus across all modes, but
• Calculations increase exponentially as nodes increase
• Further research:• To devise an integrated economic and transport model
that can account for community benefits, including wider economic benefits
14/14
Conclusions and further research
Questions?