Post on 19-Jan-2016
description
1
Economics of Road Network Ownership
1st International Conference on Transport Infrastructure FundingBanff, Canada, August 2-3, 2006
Lei Zhang
Assistant ProfessorDept. of Civil, Construction,&Environmental EngineeringOregon State Universitylei.zhang@oregonstate.edu
David Levinson
Associate ProfessorDepartment of Civil EngineeringUniversity of Minnesotalevin031@umn.edu
2
Road Network Ownership
Nationalization Privatization
Decentralization
Centralization
Competitive Market
Centralized Government
Local/Municipal Government
PrivateMonopoly
U.S. 2006
3
Questions about Network Ownership
For a specific road network…
What is the optimal ownership structure given certain
demand characteristics and cost functions?
Whether/How should ownership changes take place?
Practical Implications
Private toll roads
Public private partnership/competition
Regulation on price, investment, and ownership
4
Research Objectives
Welfare consequences of …
Centralized public ownership
Decentralized private ownership (market-oriented)
Implications of alternative ownership on
Prices (tax, toll)
Network capacity
Regulatory needs and policies
Methodological focus
Quantitative modeling
Equilibrium (point) and evolutionary (process) analyses
5
Method
Optimization
and
Simulation
Time
Year t-1 Year t Year t+1
EXOGENOUS VARIABLES Land use, Demographics, Socio-Economic Changes
Behavior: Travel Demand Estimation Zone-based, Gravity distribution, Single Mode, User equilibrium traffic assignment
Time
Technology: Maintenance Cost Construction Cost Revenue
Network t
Flow Flow Toll/Tax
MEASURES OF EFFECTIVENESS Mobility, Accessibility, Social Welfare, Financial Indicators
NETWORK MODEL
Policy: Pricing Policy Investment Policy Ownership
Network t+1
6
Centralized Ownership
Pricing Policy
Fuel taxes, Registration fees, Distance-based tolls
= Average cost pricing
Investment Strategy
Construction budget = Revenue – Maintenance Cost
Priority: Links with the highest benefit/cost ratios
Solve
for each link
At the end of each fiscal year: Budget = Expenditure
)( CapacityBCRatioMaximizeCapacity
7
Decentralized Ownership: Pricing
A Dynamic Pricing Game among All Roads
Uncertainty and incomplete information
Profit-Maximizing Pricing through Adaptive Learning
1. Estimate a demand curve
based on (price, flow) data
in previous years for each link;
2. Solve Maximize Profit(price);
3. Case A: PLow < Price* < PHigh
New Price = Price*
Case B: Price*<PLow or > PHigh
New Price = PLow(1 – j)
or = PHigh(1 + j)
A
B
Price: Toll + Monetized Travel Time
Demand: Flow
P*B PB PLow P*A= PA PHigh
8
Decentralized: Investment
Rate of Return
(Lifecycle Revenue – Lifecycle Cost)/Capital Cost
Links can borrow or earn interest through a “Bank” agent
If Rate of return > interest rate Borrow & Build capacity
Otherwise Pay off loan or Save
Profit Estimation for Capacity Expansion
Consider two sources of additional profit after expansion
1. Ability to charge higher tolls
2. Ability to attract more users
9
Fixed Demand: Equilibrium AnalysisEquilibrium Capacity on a 10by10 Grid Network
Decentralized Ownership
Socially Optimal
Centralized Ownership
10
Fixed Demand: Equilibrium Analysis 2
Equilibrium Toll
Decentralized Ownership
Socially Optimal
Centralized Ownership
11
Decentralized Ownership with Regulation
0
5
10
15
20
25
$1 $2 $3 $4 $5 $6 $7 $8 $9 Ceiling Price
Net Social Benefits
Consumers' Surplus
Profit
Socailly Optimal
Million $
Determination of the Optimal Ceiling Price
12
Fixed Demand: Evolutionary Analysis
0
5
10
15
20
25
0 10 20 30 40 50 60 70 80 90 Year
Socially OptimalDecentralized: Profit-MaximizingCentralized: Average Cost PricingDecentralized: Price Ceiling
Net Social Benefit (M$)
13
Fixed Demand: Evolutionary Analysis 2
0
2
4
6
8
10
0 10 20 30 40 50 60 70 80 90Year
Socially OptimalDecentralized: Profit-MaximizingCentralized: Average Cost PricingDecentralized: Price Ceiling
Average Toll ($)
14
Fixed Demand: Evolutionary Analysis 3
0
100
200
300
400
500
0 10 20 30 40 50 60 70 80 90Year
Socially OptimalDecentralized: Profit-MaximizingCentralized: Average Cost PricingDecentralized: Price Ceiling
Cumulative Number of Capacity Expansion Projects
15
Variable Demand: Evolutionary Analysis
0
40
80
120
160
200
0 10 20 30 40 50 60 70 80 90 Year
Socially Optimal
Decentralized: Profit-Maximizing
Centralized: Average cost pricing
Net Social Benefit (M$)
16
Variable Demand: Evolutionary Analysis 2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 10 20 30 40 50 60 70 80 90 Year
Socially Optimal
Decentralized: Profit-Maximizing
Centralized: Average cost pricing
Average Toll ($)
17
Variable Demand: Evolutionary Analysis 3
0
200
400
600
800
1000
1200
1400
0 10 20 30 40 50 60 70 80 90 Year
Socially Optimal
Decentralized: Profit-Maximizing
Centralized: Average cost pricing
Cumulative Number of Capacity Expansion Projects
18
Conclusions
Nothing is perfect
Pros Cons
Centralized Status quo Low tolls, Ineffective
Cost recovery Sub-optimal capacity
Decentralized Responsiveness High tolls
Market-oriented Risk of over-investment
When appropriate regulation is imposed (e.g. Pmax)
and/or travel demand is steadily increasing (e.g. 3%)
Results are in favor of decentralized market-oriented approach
19
Future Studies
Analysis on real-world networks (Portland, OR, Twin Cities, MN)
Consideration of hybrid ownership and ownership dynamics
Assessment of more sophisticated regulatory policies
20
Thank you!
21
Detailed Flowchart
TIME
Year t-1 Year t Year t+1
Exogenous land use, demographic, economic changes
Trip Generation
Trip Distribution
Zone production and attraction totals
UE Traffic Assignment
OD demand; k = 0
TIME
Revenue Model
Link flow k
OD cost table t + 1 OD cost table t
Cost Model
Revenue
Investment Model
Network t Network t + 1
Pricing Model
New capacity and free-flow speed
flow k = flow k–1 ?
Yes
No
k = k + 1 Link toll k
Flow, toll, travel time, OD cost
Maintenance cost Construction cost
Measures of Network Effectiveness: VHT, VKT, CS, Revenue, etc.
22
Notes:
Supply and demand do not have to be solved simultaneously; An agent-based travel forecasting system is desirable; Incremental/Adaptive changes in travel behavior are important.
Trip generation and distribution
Zone-based structure Doubly-constraint gravity model
Traffic assignment
Origin-based User Equilibrium Assignment (Bar-Gera and Boyce 2003) Generalized link travel cost function
VOT•BPR travel time + Toll
Travel Demand
)( 1 irsssrr
irs tdDnOmq
iai
a
ia
ia
aia F
fv
lt
2
11
23
Revenue and cost functions
Revenue A notion of link revenue is convenient in describing various policies
Revenue Toll • Flow
)( ia
ia
ia fE
Cost
Only a portion of maintenance cost is volume-dependent (Paterson and Archondo-Callo 1991) link maintenance cost (M) function for all links:
Empirical studies suggest (Karamalaputi and Levinson 2003) link expansion cost depends on link length, capacity and capacity change
Link expansion cost (K) function:
Length Capacity Flow
321 )()()( ia
iaa
ia fFlM
321 )()()( 1 ia
ia
iaa
ia FFFlK
Length Capacity Capacity Change
24
Model Parameters
A complete set of parameters derived for the Twin Cities
25
The agent-based simulation model of network growth seems to be a promising new approach to analyze pricing, investment strategies, and ownership structures because:
It considers both short- and long-run network dynamics;
It can evaluate alternative sub-optimal policies;
It can be easily applied to large-scale real-world networks without loss of details or computational hurdles;
It can incorporate results from studies of organizational behavior and ownership arrangements without additional modeling efforts;
Model Capabilities
26
Test Network
32 kmInitial conditions
Uniform land use with one million total trips
All links are one-lane with capacity 735 veh/hr
A very congested network (average speed 10 km/h)
Homogenous users