Chapter 1 – Introduction
2
as competition, willingness to pay tolls. The difference between these factors at the time
of the forecasts and at the opening date of the schemes may affect the traffic forecasts.
Therefore, trends of the above factors will be investigated for each country and then
related to the specific scheme.
The thesis objectives are:
1. To understand the approaches used forecasting the demand for traffic on privately
financed transport schemes;
2. To attempt to identify sources of forecast error;
3. To review the accuracy of the forecasts arising from the analytical models.
The above objectives will be met by:
1. A relevant literature review regarding forecasting for transport demand and factors
influencing transport forecasts;
2. A general review of demographic and socio-economic factors of countries interested
by the presence of the toll schemes; and by comparing the actual out turn of traffic
volumes of the schemes with the forecasted figures;
3. Running again the model on selected schemes with the update values of the
variables;
1.3 Methodology
In order to identify the key variables influencing forecast error, a number of toll
schemes are reviewed, including toll roads, shadow toll roads and public transport
schemes. The analysis is divided in two phases. The first involves a comparison of the
originally forecasted and of the actual values of factors that might be sources of
forecasts error. The differences between these are analysed by reviewing relevant
variables that are usually used within the forecast models and the assumptions made in
producing the forecasts.
For each country and for each scheme key factors such as Gross Domestic Product
(GDP), Gross National Product (GNP), Personal Disposable Income (PDI), Household
Chapter 1 – Introduction
3
Incomes, Employment, Unemployment, Population and Car Circulation and
Registration are compared with the actual figures. This phase, therefore, involves the
construction of a database of the above factors.
The database is used to draw an overview of the countries and to validate the
assumptions that have been used within the forecasts of the considered schemes.
Within the second phase a sample of significant schemes are chosen and the analysis of
the assumptions complementary to the review of the forecast models is undertaken.
Differences between the forecasted and actual values of factors that have been used
within the forecasts are found, the likely causes of over and under-prediction are found
out by running the model to the updated values. Finally, the revised (updated to actual
figures) forecasts are compared to the original forecasts and this enables to address the
causes of forecast inaccuracy to the model and/or to wrong assumptions.
In order to meet the objectives set out in paragraph 1.2, the second phase of the thesis
methodology (analytical approach, chapter 5) should be developed for a number of
schemes. Therefore, it was agreed with the collaborating company to collect data for a
minimum of eighteen schemes. Unfortunately, due to problems with the collaborating
company it was not possible to achieve the original objectives of the thesis. Data
promised and the analytical models used within the traffic forecasts of the investigated
schemes were not provided. Due to confidentiality issues, the access to data and models
was limited to the public documents. The writer has made an effort in reproducing the
analytical of two of the investigated schemes.
Therefore, the adopted approach has been a second choice in tackling the problem, and
the results depicted in chapter five and six are the best possible evaluation.
1.4 Thesis Structure
The structure of the thesis is drawn in figure 1.1 (page 5) and described below.
In chapter 1, the problem of traffic forecast for toll scheme is outlined. In this chapter,
the objectives, and the methodology of the thesis are set. The background of forecasting
demand for transport investments, factors influencing transport forecasts, toll roads,
shadow toll roads and public transport schemes is developed in chapter 2. In chapter 3
alternative methodologies are described and the chosen methodology justified. In
Chapter 1 – Introduction
4
chapter 4 the reviewed schemes are described and the demographic and socio-economic
aspects of each country interested by the scheme are depicted. Data depicted in chapter
4 are then analysed in chapter 5 where the forecast studies of two significant schemes
are reviewed in detail. This will enable the drawing of results and general conclusions.
Although the analysis of the model is undertaken only for two schemes, the discussion
of the results is placed in the context of the objectives of the study, and general
conclusions are drawn. Therefore key assumptions and models that have influenced the
forecasts to over-or-under predicting are summarised. The final chapter discusses how
the thesis objectives, set in this chapter, have been met; discusses the limitation of this
thesis; and sets future areas of research.
Chapter 1 – Introduction
5
Figure 1.1. – Structure of the thesis
CHAPTER 1
Introduction to the Thesis
CHAPTER 2
Background
CHAPTER 3
Methodology
CHAPTER 5
Data Analysis and Results
CHAPTER 6
Conclusions
CHAPTER 4
Qualitative Data
Chapter 2 – Background
6
Chapter 2 Background
2.1. Introduction
This chapter discusses the basic concepts regarding forecasting demand for toll
schemes, a description of factors usually used within forecasting models and the
meaning of various terms that are used, which may otherwise be subject of some
ambiguity. This will involve considering how economists think about the relationship
between transport and traffic demand. Complementary to this is the explanation of the
meaning of toll road, shadow toll road and public transport schemes.
2.2. Definition of Terms
From the traveller’s perspective a key aspect is the cost of using the transport system.
The cost of transport to the user is conventionally discussed in terms of generalised
cost, which includes operating costs, fares or tolls paid, incidental costs, such as parking
fees, and also the costs of time involved in making the journey. The generalised cost of
a journey will clearly depend on, among other things, the amount of congestion on the
network and may therefore vary by time of day and location. It can be expected that the
demand for transport will be inversely related to its costs as perceived by the users
(DETR, 1999).
2.2.1. Generalised cost
The following is the definition of generalised cost given by O’Flaherty (1997):
“Generalised cost” is the sum of the price of the journey and the time requirement
multiplied by the constant marginal value of time.
k
ij
k
ij
kk
ij
OPCXOVTXIVTGC
ij
++=
21
(2.1)
Where GC
ij
k
is the cost of travelling from origin i to destination j by mode k; IVT
ij
k
is
the in vehicle time required to travel from i to j by mode k; OVT
ij
k
is the total out-of-
vehicle time (e.g. walking and waiting) involved in travelling from i to j by mode k;
OPC
ij
k
is the out-of-pocket costs (e.g. tolls, fares, petrol, parking) associated with
Chapter 2 – Background
7
travelling from i to j by mode k; X
1
is the value of in-vehicle-time and X
2
is the value of
out-vehicle-time.
The concept of “Generalised cost” is strictly linked to monetary cost of travel and to the
value of time. Using generalised cost all the components that make the journey are
related in money terms.
The time components are:
• Access time (walk time);
• Waiting time;
• In vehicle time;
• Egress time (out of vehicle time).
However, according to DETR (1999), generalised cost varies by mode.
For cars, generalised cost is a combination of:
• In-vehicle travel time;
• Operating costs (related to distance travelled);
• Parking cost;
• Tolls or congestion charges.
For goods vehicles, the components are similar, except that different vehicle operating
costs and values of time are used.
For public transport users, generalised cost is a combination of:
• Walking time from the origin to a stop or station;
• Waiting time for the service;
• Fare;
• In-vehicle time;
• Penalty representing the inconvenience of changing between services;
• Walking time to the destination.
2.2.2. Value of time
According to DETR (1999), resource values measure actual resources consumed in
travelling during working time and are related to market wage rates. For non-working
time, no market exists and the values reflect people's willingness to trade time for
money.
Chapter 2 – Background
8
2.3. Forecasting Traffic Demand for Transport Investments
The Private sector is more often than not involved in building and operating
transportation network at its own expenses and risk, in return it receives the revenue
from road toll charge.
The risk of toll scheme can be classified in four categories: construction, financial,
management and forecast. One of the most important issues concerning transport
schemes is to forecast the most realistic traffic demand, which in turn has a direct effect
on the revenues of the private investor.
Demand for transport
“The demand for transport is a derived one, and each journey is unique in time and
space…” (Stubbs et Alt., 1986).
It is well accepted by economists, such as Cole (1998), Stubbs et Al. (1986), DETR
study (1999), that demand for transport is very largely a derived demand arising from
other activities.
According to DETR (1999) the demand for freight transport arises from the following:
• The purchase of goods and services by final users which requires the carriage of
inputs to the place of production; and
• The distribution of products from the place of production to the final point of use via
the point of sale.
DETR (1999) study argues, also, that the demand for passenger transport arises from the
following:
• Journeys that bring people to work, education and training, in economic terms
supplying labour to production;
• Journeys that allow access for individuals to consumption opportunities, such as
shopping, leisure and tourism;
• Journeys that allow access for individuals to other individuals (e.g. visiting friends
and relatives); and
• Journeys that provide direct value to individuals (e.g. travelling on a preserved
steam railway).
Chapter 2 – Background
9
As the economy grows and production, sales and incomes rise, these various demands
for transport can all generally be expected to increase (DETR, 1999).
Once, the general concepts regarding traffic demand and user’s cost perception have
been explained, the relationship between travel demand and cost-per-trip can be
introduced.
Forecasts are usually produced for two different scenarios: a base case or do-minimum
case and a with scheme case.
Figure 2.1 shows the demand/supply relationship in a cost per trip/travel demand
diagram. In this diagram two curves are drawn; the demand curve; and the supply curve.
Point A represents the equilibrium point and this scenario can be defined as the base
case scenario.
Figure 2.1. – Demand/Supply Curve (source: DETR, 1999).
If an intervention reduces road capacity or increases travel costs (by increasing toll
charge), the supply curve for the with-scheme case will lie above the base case curve.
Conversely, if the intervention increases road capacity or decreases costs (road
improvements or a decrease in toll charge), the new supply curve will lie below the base
case curve. The equilibrium point will be different, thus the traffic demand will change
(point B fig. 2.2).
Chapter 2 – Background
10
Figure 2.2. – Demand/Supply Curve base case (source: DETR, 1999).
Changes in external factors such as economic growth, household’s income, car
ownership population, households, employment, and other factors may change the level
of demand shifting the demand curve as shown in figure 2.3. This curve represents all
the responses that travellers may make to changes in travel cost.
The task within forecasts appraisal is to find the new equilibrium point D. This point
represents the intersection between the demand curve shifted by external factors and the
supply curve in the with-scheme case (fig. 2.3). The shaded area C
1
C
2
CD shown in
figure 2.3 represents the transport benefit.
Investments requires a consideration of the future, because the expenditure are always
in the short term (three to five years), while benefits in terms of revenues are in the long
term period (10 to 25 years).
Chapter 2 – Background
11
Figure 2.3. – Demand/Supply Curve forecast case (source: DETR, 1999).
According to DETR (1999), in order to produce accurate traffic forecasts, the following
requirements are very important.
• The relative slopes of the demand and supply curves;
• The accuracy with which the shift in the demand curve can be forecast over time;
• The accuracy with which the change in transport costs can be predicted by the
supply curve;
• The accuracy with which the equilibrium positions can be found.
At the beginning of the chapter it was outlined that demand for transport is a derived
demand and that it is a combination of various elements. Therefore, in order to produce
accurate traffic forecasts, these various elements should be identified. These elements
have been studied for years and by several researchers.
Dunphy and Fisher (1996) have studied the relationship between urban densities, socio-
economics characteristics of residents and their travel characteristics. They found that
household travel, as measured in miles per household, increases with income.
Kockelman (1997) has developed a model based on several variables including, land-
use, accessibility, vehicle miles travelled per household, non-work home based per
Chapter 2 – Background
12
household, auto ownership and mode-choice. It was found that land-use and
accessibility were more relevant to travel behaviour.
Cevero (1996) examined the influence of mixed land-use and built environment on the
mode used for commuting, the commuting distance and household vehicle ownership.
The same author concluded that mixed land-use has a strong impact on all commuting
mode-choices and that neighbourhood density and mixed land-uses affected vehicle
ownership.
He also investigated how the built environment affects mode of access to and from rail
stations and the size of the station catchment area. The built environment was found to
influence choice between rail and drive.
Badoe and Miller (2000) have reviewed several studies regarding the relation between
transportation and land-use. As a result of the review they have identified which factors
are influencing travel demand and the relationship between them (fig.2.4).
Figure 2.4. – Factors influencing travel demand (source: Badoe and Miller, 2000).
Accessibility
Auto Ownership
Transit service
Socio-economics
Road network
Demographics
Residential Density
Employment
Density
Travel
Demand
Neighborhood
design
Chapter 2 – Background
13
2.4. Factors Influencing Transport Forecasts
Economists (Cole, 1998) argue that forecasts demand are based on three general
elements (fig.2.5):
• Land use, which relates to the reason of the journey and its origin and destination;
• Travel costs, which is related to the concept of generalised cost. This is determinant
for the decision to undertake or not the journey and if it is undertaken by which
mode of transport;
• Economic Factors. As it has been said demand is a derived demand and its level of
activity depends in general by the level of output of the economy. However, there
are some cases in which travel is not governed by economic factors, such as journey
to school.
Figure 2.5. – Economic Model for Transport Forecasts.
The three main elements involved in forecasting models for transport are strictly related.
Changes in one of more of these elements will influence the others and as a result the
forecasts for traffic demand. Moreover the relative importance of these three elements
depends on the time-scale of the forecasts; in the short term behaviour in determining
travel or movement is influenced mainly by user cost; in the long term fundamental
changes in land use will alter the whole pattern of travel demand (Cole, 1998).
Several variables are involved in traffic forecasts, they can be grouped in two main
groups: demographic and socio-economic variables. Due to the number of variables
involved in traffic forecast studies, assumptions on the future growth of these variables
are usually made.
LAND USE
ECONOMIC
FACTORS
TRAVEL COSTS
FORECASTS
Chapter 2 – Background
14
2.4.1. Demographics
It is well accepted by economists that, from a macroeconomic perspective, demographic
variable such as the growth and distribution of population, age and sex of household
have a strong impact on traffic demand. Uncertainties in producing forecasts for these
variables are cause of inaccuracy in predicting traffic volume for transport schemes.
2.4.2. Socio-Economics
“Generally, socio-economic variables, particularly income, car ownership and employed
persons, have been found to be the main factors affecting travel demand.” (Stubbs et
Al., 1986).
According to the above assertion it has been found in literature that socio-economic
factors such as car ownership, household incomes and size, employment, Personal
Disposable Income (PDI), Gross Domestic Product (GDP) are considered to be the most
important in evaluating traffic demand forecasts.
Once again uncertainties in the forecasts of these variables produce the effect of
inaccuracy in traffic forecasts.
2.5. Transport Forecasts Models
In figure 2.6 the structure of the classic transport model is depicted. This is presented as
a sequence of four sub-models: trip generation, distribution, modal split and assignment.
The objective of the paragraph is to focus on the “trip generation” sub-model and on its
analytical aspect.
According to the definition of trip generation given by Ortuzar and Willumsen (1994),
trip generation sub-model includes all the data that are used to estimate the total number
of trips generated and attracted by each zone of the study area.
Chapter 2 – Background
15
Figure 2.6. – The classic four-stage transport model (source: J.G. Towriss, 2000)
The model used in evaluating trip generation is usually a Multiple Linear Regression.
The task is to find a linear relationship between the number of trips produced or
attracted by each zone and average demographic and socio-economic characteristics of
the household in each zone. The analytical formulation of the model is expressed by
equation 2.2.
Y
i
= a + bX
1
+ cX
2
+…+ kX
k
+ E
i
(2.2)
Where:
X
i
…X
k
represent the demographic and socio-economic variables
a represents the model constant;
b…k are the model coefficients;
E
i
is the model error.
Zones
Network
Base-year
Data
Future
Database
Base year Future
Trip Generation
Trip Distribution
Modal Split
Assignment
Evaluation
Ricevi informazioni sui nostri servizi, sulle offerte e non perdere news e consigli su università e lavoro.
