Vous êtes sur la page 1sur 4

Transportation Research Part D 13 (2008) 347350

Contents lists available at ScienceDirect

Transportation Research Part D


journal homepage: www.elsevier.com/locate/trd

Notes and comments

Choice valuation of trafc restrictions: Noise, pollution,


and congestion preferences. A note
Carlos Pestana Barros a, Peter U.C. Dieke b,*
a
b

Instituto Superior de Economia e Gesto, Technical University of Lisbon, Rua Miguel Lupi, 20, 1249-078 Lisbon, Portugal
School of Recreation, Health, and Tourism, George Mason University Manassas, Virginia 20110, USA

a r t i c l e
Keywords:
Pollution
Lisbon
Mixed logit model
Public policy

i n f o

a b s t r a c t
This Note looks at the choice valuation restrictions on trafc entering the city of Lisbon,
based on individual preferences in relation to noise, pollution and congestion. The analysis
employs a questionnaire distributed in 2007 to ascertain the signicant characteristics of
traveling to Lisbon, with the aim of curbing the number of cars that are associated with
the probability of individuals supporting a charge on motor vehicles entering the city.
The model also takes into account the uncontrolled heterogeneity of the data.
Published by Elsevier Ltd.

1. Introduction
Various cities have introduced a congestion charge in an attempt to reduce the number of motor vehicles entering citycenters; Singapore (1976), Rome (1998 and extended in 2001), Durham (2002), London (2003 and extended in 2007), Stockholm (2007), and Valletta (mid-1990s). It is also under consideration at many more cities including Lisbon. Here we look at
the behavioral process that leads Greater Lisbon citizens to evaluate the negative externalities of trafc and support the
imposition of a charge on trafc entering the city.
2. Context
Lisbon is an old city, with many buildings in a poor condition because of a rent policy that for more than 50 years limited
rent increases leading to owners neglecting maintenance and refurbishment. As a result many families relocated to the suburbs and commute into the city for work. The result has been high levels of noise, pollution and congestion, in addition to
severe parking problems in central Lisbon. While no policy to restrict the trafc has formerly been announced, in 2007, the
municipal council implemented a more rigorous enforcement of city-center speed limits using radar and closed circuit television on principal arteries. This does little, however, to reduce congestion, pollution and noise, because it fails to address the
number of vehicles. The media has assumed the role of raising awareness of the problem focusing on policies to transfer
commuting by car with bus or train journeys.
Whereas Lisbon is reasonably well served by its publicly-owned bus and metro networks, congestion is fuelled by the
daily inux from the outer suburbs, principally along four major highways, one secondary highway and one coastal dual-carriageway. Although the local railway networks offer services from four directions into Lisbon, three of these pre-dating the
rapid development of the satellite suburban areas.

* Corresponding author. Tel.: +1 703 993 4260; fax: +1 703 993 2025.
E-mail address: pdieke@gmu.edu (P.U.C. Dieke).
1361-9209/$ - see front matter Published by Elsevier Ltd.
doi:10.1016/j.trd.2008.03.006

348

C.P. Barros, P.U.C. Dieke / Transportation Research Part D 13 (2008) 347350

Moreover, the river Tagus is a major transport barrier. Lisbon stands on its north side, while many thousands of workers
live to the south in far cheaper districts or new towns. Until the inauguration in 1998 of a second major bridge some distance
upriver and the opening in 1999 of a privately-owned rail link beneath the only road bridge into central Lisbon, workers from
a wide radius to the south were obliged to cross the river by road and/or ferry. In Portugal, charges are levied only on interurban highways. Thus, the country has no experience with urban congestion charging. Many citizens, however, are aware of
Londons scheme and various politicians have expressing support of a similar policy for Lisbon and the second city, Porto.
3. Research design
We consider individuals who travel to Lisbon by car and declare that support for restrictions on private cars entering the
city. There are a number of ways to look at the factors inuencing this support:
 Individual commuters who choose to support the charge are those who are attracted by choice attributes related to pollution, noise and congestion. This is a traditional hypothesis in transportation demand models (Hensher and Puckett,
2007; Sarova et al., 2007).
 A commuter who chooses to support the charge is characterized by socio-economic characteristics regarding, age, income,
level of education, social class, married and with children (Srinivasan and Athuru, 2005).
 Travelers who choose to support the charge are those who have been adequately informed. Perception formation derives
from information that helps travelers clarify and evaluate travel decisions.
 Travelers who choose to support the charge are those who have temporal constraints and thus less time to want to spend
on travel (Espino et al., 2006).
To examine these alternatives, a mixed logit representation is used that assumes the probability of supporting the charge,
can be described by a logit-probability function of the exogenous variables (Train, 2003). The probability for choice is as:
Z 1
PrChoicei jvi
Pb; vi Nb5 jl5 ; r5 db5
1
1

where N() is the normal distribution, and


vi b0 b1 Pollution b2 Noise b3 Congestion b4 Income b5 Civilstate b6 Class b7 Education b8 Group
b9 Newspaper b10 TV b11 Treserve ei
Variables are determined using a questionnaire. Variable vi is measured by the probability that an individual declares that he
supports the charge taking a value of unity for a positive response, and measure Xi as observed characteristics.
4. Empirical analysis
To gather information, the questionnaire was administered to a stratied, random sample travelers entering Lisbon during the morning rush-hour.1 The sample was stratied by transport mode (car, bus, train and metro), using the traveler database available in the National Trafc Institute. Three thousand questionnaires were administered. The interviewers approached
the passenger on the chosen modes, using a random procedure, in which the questionnaire was distributed by hand, together
with a stamped envelope to return the completed questionnaire. This procedure resulted in the return by mail of 1652 completed questionnaires (Dillman, 1978).
The rate response does not differ signicantly among the population by age variable (v2 = 8.53, p = 0.05) or gender
2
(v = 7.55, p = 0.05) indication the respondents are representative of Lisbon travelers. The general characteristics of respondents were male (52%), with an average age of 45 and educated to college degree level; for more details see Table 1.
The survey comprised three sections: the rst, characterizes the commuters in relation to their socio-economic prole
and included variables such as social class, age, gender, family and education; the second includes the tripographic variables
and constraints such as budget, travel frequency, and temporal constraints; and the third, their preferences with regard to
noise, pollution and congestion. A seven-point Likert-type scale assesses the policy attributes ranging from without importance (1) to very important (7). An open-ended question was asked on the willingness-to-pay a congestion charge that
ranged from zero to 10, with an average value of 5 and a standard deviation of 4.8. Table 1 shows the signicant variables.
The mixed logit results display evidence of unobserved heterogeneity, depending on income and education, common to
the three segments analyzed. The survey also shows heterogeneity in the effect of noise on vehicle drivers, inducing the estimation of a mixed logit model with random coefcients for these variables. The nal results are seen in Table 2.
The overall t of this model is reasonably good, with a Chi-square statistic square value of 19.5 with 41 degrees of
freedom and level of signicance of 0.0067. The other models display similar values. The variables are, for the most part, sta-

1
The questionnaire was pre-tested on students of economics at the Technical University of Lisbon and following the administration of the nal survey, a
stratied random sub-set of 50 respondents were contacted by phone see if any problems had arisen in completing the questionnaire; none were detected.

349

C.P. Barros, P.U.C. Dieke / Transportation Research Part D 13 (2008) 347350


Table 1
Characterization of the variables
Variable

Description

Mina

Charge

Do you favor a charge to restrict cars from entering the city: yes = 1, No = 0
Destination attributes hypothesis
What was the importance of pollution in your decision? (1-without importance; 7-extremely
important)
What was the importance of noise in your decision? (1-without importance; 7-extremely
important)
What was the importance of congestion in your decision? (1-without importance; 7-extremely
important)
Socio-demographic characteristics hypothesis
Commuter monthly income (1-lower than 1000 ; 5- equal to or higher than 5000 )
Marital status (1-single; 2-married; 3-with children)
Social class (1-lower; 2-middle; 3-upper-middle)
Education (number of years of education) (4-primary school, 24-Doctorate)
Travel in groups
Information hypothesis
What was the importance of information obtained in newspapers in your decision? (1-without
importance; 7-extremely important)
What was the importance of TV information in your decision? (1-without importance;
7-extremely important)
Temporal constraints hypothesis
How long is your journey? (1-less than 15 min; 215 min or more, but less than 30 min; 3-more
than 30 min; 4-More than 45 min; 5-more than an hour)

0.490

5.454

1.465

5.923

1.368

5.470

1.532

1
1
1
4
0

5
3
3
24
1

1.981
2.260
2.124
14.013
0.252

1.140
1.451
0.911
3.461
?

4.766

1.524

4.038

1.483

4.036

1.018

Pollution
Noise
Congestion

Income
Civilstate
Class
Education
Group
Newspaper
TV

Long
a
b

Maxb

Mean

Std.
Dev.

MinMinimum.
MaxMaximum.

Table 2
Mixed logit parameter estimates and t-statistics (dependent variable: Do you favor a charge on cars entering Lisbon?: Yes = 1, No = 0)
Variables

Intercept
Pollution

Mixed logit segment: Drivers

Mixed logit segment: train commuters

Mixed logit segment: bus commuters

Coefcients (t-ratio)

Coefcients (t-ratio)

Coefcients (t-ratio)

2.498
(1.882)***
0.163
(3.281)*

0.174
(3.229)*

0.204
(2.188)**
0.427
(2.679)*

0.083
(2.001)*
0.086
(0.862)
0.152
(0.985)
0.196
(2.397)*

3.212
(21.089)*
0.132
(3.930)*
0.051
(2.836)*
0.213
(2.934)*

0.317
(2.955)**
0.512
(3.658)*

0.052
(2.712)*
0.512
(0.743)
0.012
(3.201)
0.135
(2.249)*

4.135
(11.473)*
0.218
(4.498)*
0.412
(3.844)*
0.351
(3.209)*

0.128
(2.928)**
0.612
(2.407)**

0.051
(1.804)
0.213
(1.862)***
0.123
(0.820)
0.035
(2.212)**

0.379 (2.605)*
0.255
(2.036)**
0.032 (3.113)
642
294.75

0.315
(2.030)
0.051 (2.273)
568
295.316

0.591
(3.969)**
0.125 (3.227)
442
299.967

Noise
Congestion
Income
Civil state
Class
Education
Group
Newspaper
TV
Long
Random effects
Noise
Income
Education
Observations
LogLikelihood

Dependent variable: choosing to restrict cars entering the city.


Means 10%.
Means statistically signicant at 5%.
***
Means statistically signicant at 1%.
*

**

350

C.P. Barros, P.U.C. Dieke / Transportation Research Part D 13 (2008) 347350

tistically signicant and three heterogeneous variables are identied. Two are common to all travel mode types, education
and income, while noise is specic to drivers.
The ndings pointed to a signicant correlation between the probability of supporting a charge on cars and the exogenous
variables. For the mixed logit framework, the probability of supporting a charge increases with most of the variables but support decreases with groups traveling in cars and with long-distance travel in all modes of transport.
Since the three environmental variables are positively related with the probability of supporting the charge and are statistically signicant there is some support for the notion that acceptance of a charge is related to environmental concerns.
However, noise is random and therefore reects heterogeneous opinion among the car drivers. Equally, the socio-economic
variables are, for the most part, positive, although income and education, class and civil status are also positive and statistically signicant and homogenous. However, the variable group is negative for cars. This mixture of results suggests that
socio-economics are not consistent forces affecting support for congestion charging. Regarding knowledge and information,
TV and Newspaper are both positive and statistically signicant variables indicating this is an important affect but the negative signs for long-distance commuting indicate that temporal constraints are not.
5. Conclusions
These results suggest that a high level of support for trafc restrictions exists among afuent travelers and that commuters are highly sensitive to the effects of pollution, noise and congestion, although less so to noise. Environmental awareness
is supported by the information obtained from newspapers and television. The length of travel negatively affects support for
a charge. Thus on average, commuters support a charge for entry into Lisbon, but car drivers who travel in groups are
opposed to it, as are persons with long daily journeys in all modes of transport.
References
Dillman, D. (Ed.), 1978. Mail and Telephone Surveys: The Total Design Method. Wiley, New York.
Espino, R., Romn, C., Ortuzar, J de D., 2006. Analysing demand for suburban trips: A mixed RP/SP model with latent variables and interaction effects.
Transportation 33, 241261.
Hensher, D.A., Puckett, S.M., 2007. Congestion and variable user charging as an effective travel demand management instrument. Transportation Research
Part A 41, 615626.
Sarova, E., Gillingham, K., Houde, S., 2007. Measuring marginal congestion costs of urban transportation: do networks matter? Transportation Research
Part A 41, 734749.
Srinivasan, K.K., Athuru, S.R., 2005. Analysis of within-household effects and between-household differences in maintenance activity allocation.
Transportation 32, 495521.
Train, K., 2003. Discrete Choice Methods with Simulation. Cambridge University Press, Cambridge.

Vous aimerez peut-être aussi