Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Proceedings of the 11th International IEEE

Conference on Intelligent Transportation Systems


Beijing, China, October 12-15, 2008

A Comparative Study of Parking and Congestion Charge Policies


based on Transport Mode Choice Estimation
S. W. Feng, J. J. Ye

Abstract— With more steps striding towards the practice of some trial on the other marketing tools of TDM like the
market tools like parking charge and congestion charge policies, parking charge in central region. In 2007, the Municipal
much attention has been paid to the pricing and policy Government of Shanghai and Shenzhen expressed their
performance. Based on the survey with willingness-to-pay programs to ease the traffic congestion through congestion
measures to commuters in Shanghai, some estimation of binary
charge planning. Actually according to the Shanghai
Logit model has been carried out to reveal the common and
special factors acting on mode choice. Two Scenarios ‘without ’ Metropolitan Transport White Paper (Municipal Government
and ‘with’ charge polices are designed to reveal the difference of Shanghai, 2002), some charging policies like parking
between all informants and commuters with car-ownership, and charge and congestion charge have been designed step by step
further to find out special explaining variables to different to lighten the traffic pressure stemming from the high speed
policies. The analysis shows that the transfer of congestion of current motorization and urbanization. Around 2006, a rise
charge from car users to their units is one of the hardest
of parking charge in central business district (CBD) resulted
obstacles to mode change when implementing the congestion
charge policy in China’s cities. in some progress to keep down the usage of private vehicles
in Shanghai and the city is striding forward to more integrated
I. INTRODUCTION policies like public transport priority and the congestion
charge.
A MONG all transportation demand management (TDM)
measures, the parking charge and congestion charge
would be the most attractive and effective marketing tools to
The core questions to implement the market tools of TDM
lie in how to determine the price mechanism and how to
predict or evaluate the performance before/after these policies
transportation economists. Looking back to the experience of
being adopted. Though a stable theoretical foundation of road
congestion management practice of different cities in the
pricing has been built by Arnott (1993), Verhoef (1996),
world, various public policies, including parking
Yang & Huang (1998), since the early researchers of
management, road pricing and public transport priority, have
marginal pricing theory, like Pigou (1920), Walters (1961)
been carried out according to a before-and-after series and it
and Vickrey (1969), the evaluation of the performance of
seems that these polices have been integrated together for
such policies has still been an ambitious and inconceivable
achieving the comprehensive effects to change the travelers
idea to policy-makers and traffic engineers because they must
behavior, which is the final target of TDM. Generally
encounter and deal with the complicated transportation
speaking, those countries and regions that adopted the
system and the complex traveling behavior. Some (Li, 2002)
marketing tools of TDM left a trail of first using parking
had been achieved on the transportation engineering side due
charge due to its low cost and easily operating, and then
to the measurable parameters of traffic flow status comparing
testing the congestion charge to further strengthen the
before and after the policy implement. The prediction and
effectiveness of traveling behavior management by
analysis of TDM policy performance could not be
increasing the car-based expense, when people travel across a
accomplished without the mode choice observation and road
certain region or during a certain time period (Hau, 1990;
users’ attitude towards the policies. Some studies (Axhausen,
Small, 1998; Chin, 2002; Litman, 2006).
1991; Jaccard, 2006) evolved from the demand side by setting
Inspired by the successful practice of congestion charging
up transport behavior survey and mode choice model and this
in some countries and cities, like Singapore, London and
paper will go further along this path.
Stockholm, a few China cities recently pay great attention to
In order to predict and evaluate the performance of
congestion charging regulation after having suffered bitterly
marketing tools of TDM, especially parking charge and
from the serious traffic congestion for decades and have made
congestion charge, this paper conducted a survey involving
more than 1000 commuters in Shanghai to obtain the
Manuscript received May 31, 2008. This work was supported in part by fundamental data for mode choice setting. Moreover, based
the National Natural Science Foundation of China under Grant 70571048.
S. W. Feng is with the Public Management Department, the School of
on a contingent valuation method (CVM), the commuters’
Public Economics and Administration, Shanghai University of Finance and willingness-to-pay (WTP) to charging policies was contained
Economics, 777 Guoding Rd., Shanghai, P.R. China. (Phone: in the questionnaire to collect their reaction and attitude as an
86-21-65904405; e-mail: fsuwei@mail.shufe.edu.cn)
J. J. Ye is with the Public Management Department, the School of Public
input of policy variables. Finally, the binary Logit model
Economics and Administration, Shanghai University of Finance and (BLM) was estimated to find out the coefficients that impact
Economics, 777 Guoding Rd., Shanghai, P.R. China. (E-mail: the mode change. By comparing different charging policies,
ye.1103@yahoo.com.cn)

1-4244-2112-1/08/$20.00 ©2008 IEEE 506


some important factors to mode choice were filtered for is one kind of stated preference methods that is widely used to
giving some advice in the policy-making and evaluate the public goods or service that are sometimes
price-determining process. difficult to give a price through market (Hanemann, 1994).
The following content is arranged in this way: a brief By asking directly the maximum willingness to pay (WTP) of
introduction of involved methods like BLM and WTP obtaining a certain conceived good or service or the minimum
measurement appears in section II and the survey information willingness to accept (WTA) of losing such good or service,
can be found in section III. Further, while the estimation of CVM can evaluate the general usable and non-usable
BLM coefficients, two scenarios ‘Without Policies’ and ‘With economic value of public goods or service by summing the
Policies’ are designed to first compare the mode choice of separate answers together.
different commuter catalogs and then to compare the A pay card measure is introduced in this paper to obtain the
performance of two charging policies. Some insights from the subjective valuation of informants when the general WTP is
survey analysis and model estimation have also been given in needed to estimate the price of certain charging policy. For
section IV to highlight the implementation of the charging example, for the parking price, the question is: What is the
policies in China’s cities. maximum charge you could pay by yourself for parking your
car per day at work? And there are 8 options varying from ¥5
II. METHODOLOGY to ¥100 with different intervals which the informants can
The traditional method to study travel behavior is the choice. In this way, the distribution of WTP can be obtained
discrete choice model (see, Train, 2002, for a review of and further the relationship between the number of certain
theories and methods) since the mode choice is usually mode choice commuters and certain policy charging can be
separated into two catalogs by using the private transport tool connected by statistics, which serve as the mode choice
or not. Let p i be the probability of traveling by car, probability and the input of policy variable in the BLM
model.
1 − p i be the probability of choosing other modes, a binary
Logit model (BLM) can be described as III. SURVEYS AND COEFICIENTS ESTIMATION
pi (1) The individual decision of mode choice involves many
= e Zi
1 − pi social-economical factors such as mode preference, car
Further ownership, trip purpose, transport expenses, congestion
z i = β 0 + β 1 x1 + β 2 x 2 + " + β n x n + ε i (2) degree, value of time, the availability and accessibility of
Where β j ( j = 0, " , n ) are the coefficients to be estimated, public transport, and so on. Analyzing these factors on mode
choice and further evaluating the performance of charging
ε i is the stochastic term with the hypothesis of Gumbel policies, a survey needs to be conducted to supply the
distribution and x j ( j = 1, " , n ) are the factors to explain demography and mode data and to reflect how travelers
behave under different parking and congestion charge.
the mode choice, including the social-economical factors and
For the purpose of understanding the price mechanism to
policy variables. This BLM has successfully used in many
mode choice and evaluating the effect of charging policies,
studies (Axhausen, 1991; Jaccard, 2006) concentrating on
the survey conducted in this paper concentrates on
traffic behavior and it is also applied in this paper to describe
commuters who travel day-by-day between home and work
the commuters’ mode choice whether commuting by car
place in the most congested rush hours in Shanghai. Two
(including car-pool) or other modes (like bus, metro, bicycle
rounds of survey have been conducted to collect enough
or foot).
behavior data to support the mode choice model and also to
The charging price is a sensitive factor while using market
supply the variables of policies when analyzing the
tools of TDM. How to determine the price is also one of the
performance. The survey was carried out from late 2007 to
key issues to policy-makers because there is no comparative
early 2008 and 1068 commuters in Shanghai were involved.
point of price among different cities where the charging
Finally 962 valid samples were obtained to meet the demand
policies have been implemented or have been planning to be
of the model estimation and policies analysis.
carried out. If the price would be merely determined through
The survey data reveals three catalogues of information:
administrative regulation, the change of mode choice could
(1) Fundamental demographical information of the
be sometimes hard to estimate and the policy performance
commuters, including gender, age, institutional affiliation
would not be predicted precisely either. Actually before
(unit), personal and family income and number of family
policy implementation, some information from the road users
members. (2) Commuting information, which contains
is needed if the prediction of policy performance is required
ownership of private car, places of home and work,
by corresponding governmental department, or even by the
commuting mode and time, commuting expense according to
public.
three cases: without a car, with car(s) but no use, and using
The contingent valuation method (CVM) can satisfy the
car for commuting. Finally the ratio and distribution of
demand of such information from the road users’ side. CVM

507
commuting expense reimbursement by affiliation is also ‘Parking expense’, are also significant to mode choice.
inquired. (3) Attitude and WTP to charging policies: Based on the WTP measures in the survey, Scenario two
consisting of WTP to parking (if it would be rising) or considers ‘With Policies’ situation and supposes to the
congestion charging (if it would be conducted), awareness commuters with car-ownership: (1) the parking charge would
media of policies, attitude and acceptance to policies, increase (shown as different options in WTP questions) and
personal assessment to the effectiveness of single and dual (2) the congestion charge would be implemented respectively
policies, and so on. or dually with parking policy.
With these data, two levels of assessment to the In the estimation, if the charge sets on ¥10, those who
performance of parking and congestion charge policies have expressed their WTP less than ¥10 would be supposed to
been carried out. First, if dividing the informants into two change their commuting mode from car driving to others.
groups according to the ownership of private car, a
TABLE I
comparison of their WTP and attitudes towards the policies COMPARISON OF ALL INFORMANTS AND COMMUTERS WITH
can be done to reveal the social difference and mode choice CAR-OWNERSHIP FROM COEFFICIENTS
preference between two groups. Second, an empirical model Commuters with
Variable All informants
based on discrete mode choice can be set up to evaluate and car-ownership
show the relationship between the WTP charge and the mode Gender 1.0757 (0.2623)* 1.0737 (0.2880)*
choice, which will give a glimpse of what affects the car using Personal 0.0381 0.0347
income (0.0179)** (0.0189)***
in nowadays fast motorization and what level of pricing could Work place 0.6199 (0.2582)** Not significant
be adopted to change commuting mode efficiently. Car-ownership 3.4354 (0.3156)* Not significant
For these purpose, two comparative scenarios have also Commuting -0.0035 -0.0059
time (0.0027)** (0.0031)***
been set up to evaluate the undertaking parking charge policy Other expense 0.0012 (0.0004)* 0.0016 (0.0005)*
and the planning congestion policy in Shanghai. Scenario one Parking 0.0176 0.0162
is ‘Without Policies’ which serves as a background study. It expense (0.0035)* (0.0034)*
Constant -4.3515 -0.4751
supposes: (1) the parking charge would not rise but remain (0.4483)* (0.3425)***
ordinary level, and (2) there would be no congestion charge Notes: Standard errors in parentheses.
policies. Since the ownership of a car has great impact on car *, **, *** indicates significance at the 1%, 5% and 10% levels.
using, the samples are analyzed in two groups: all informants
and commuters with car-ownership for comparison. The Then by calculating the probability of commuting by car and
BLM coefficients have been estimated with given probability the parking or congestion charge as input variable, the BLM
of commuting by car and the explaining variables information coefficients can be estimated and correspondingly the
from the survey. For all informants, equation (2) can be equation for parking charge policy is
described in detail as z i = − 7 .6687 + 1 .1835 x gender + 0 .0392 x P _ income
z i = − 4 .3515 + 1 .0757 x gender + 0 .0381 x P _ income
− 0 .0085 x time + 0 .0015 x other + 0 .0366 x parking
+ 0 .6199 x workplace + 3 .4354 x car − 0 .0035 x time (3) _ exp _ WTP

(5)
+ 0 .0012 x other _ exp + 0 .0176 x parking For congestion charge policy,
For commuters with car-ownership, we have z i = − 7 .6505 + 1 .4320 x gender + 0 .317 x unit
z i = − 0 .4751 + 1 .0737 x gender + 0 .0348 x P _ income − 0 .0098 x time + 0 .0015 x other + 0 .0344 x congestion
(4) _ exp _ WTP
− 0 .0059 x time + 0 .0016 x other _ exp + 0 .0162 x parking (6)
Table I compares the coefficients of all informants with the Table II shows the coefficients comparison of two policies.
commuters with car-ownership. Comparing to all informants, The explaining variables and their significance release what
the sample of commuters with car-ownership losses two kind of factors may affect commuters’ mode choice when
explaining variables as the ‘Work place’ and the either parking charge or congestion charge policy would be
‘Car-ownership’. It is easy to explain that the situation of carried out. The following are the key points when comparing
work place from home will determine commuters’ choice of these two policies:
using car or not and a commuter tends to use his/her car to 1) Common variables: ‘Gender’, ‘Commuting time’,
work when he/she owns a car. For the common variables of ‘Other expense’ and ‘Constant’ remain the same as Scenario
two samples, ‘Gender’ implies that male commuters prefer one, which shows that the gender and commuting cost
using car than female; ‘Personal income’ relates to have consisting of time and money are the common factors acting
sufficient money covering the expense of using a car, so that efficiently when commuters make decision of mode choice.
income is an important and determinant factor to own and use 2) Special variables to ‘Parking’ policy: They are
a car for commuting. ‘Commuting time’ usually links to far ‘Personal income’ and ‘Parking charge’ (in WTP form). The
distance separating from home to work and congestion parking charge in Shanghai has risen since 2006, so that the
transportation that are negative impacts for driving to work. commuters have had common sense of parking management
Expenses of car use, covering gasoline, usage tax and fee, and and are easy to react and accept the policy when the charge is

508
supposed to increase continuously in the survey scene. It is long march of Shanghai Manipulate Government who has
clear that the factor of ‘Personal income’ still remains as same been in the battle to the traffic congestion for decades. The
as Scenario one while the ‘Parking charge’ will has positive survey to more than 1000 commuters supplies input data of
correlation to choice of car as commuting tool. commuter demography, traveling preferences, attitude
3) Special variables to ‘Congestion’ policy: They are towards price levels of policies which is answered by WTP
‘Unit’ and ‘Congestion charge’ (in WTP form). The variable pay card questions. With these data, different coefficients of
of ‘Unit’ is divided into five catalogs in survey: (1) BLM are estimated to describe the important factors acting on
State-owned enterprises, (2) institutions, (3) governmental policies. Further analysis combining the survey and the
estimation shows:
TABLE II
COMPARISON OF PARKING CHARGE AND CONGESTION CHARGE FROM 1) Both the ‘Parking’ and ‘Congestion’ policies could be
COEFFICIENTS effective to mode change of commuters from driving car to
Congestion charge public transport. In the survey, 67.1% of informants will
Variable Parking charge policy
policy abandon driving car to work when the parking price rises
Gender 1.1835 (0.3057)* 1.4320 (0.3523)* higher than ¥10 while 79.6% will change their driving mode
Personal 0.0391 (0.0201)** Not significant to others when the congestion charge reaches higher than ¥10.
income
Unit Not significant 0.3176 (0.1487)** In this way, the mechanism of how market tools affect
Commuting -0.0085 -0.0098 commuters’ mode choice can be observed and scaled.
time (0.0035)** (0.0039)** 2) The ‘Congestion’ policy has more effective reaction than
Other expense 0.0015 (0.0004)* 0.0015 (0.0004)*
Parking charge 0.0366 (0.0071)* - the ‘Parking’ policy. The difference of commuters’ mode
(WTP) change between two policies reaches 12.5% under the same
Congestion - 0.0344 (0.0049)* rise of prices to ¥10, which shows that the ‘Congestion’
charge (WTP)
Constant -7.6687 (1.6760)* -7.6505 (1.2874)* policy will work better than ‘Parking’ policy in the mode
change by market tools. This has already gotten some
Notes: Other expense relates to the commuting cost except parking.
practical evidence from other countries and cities in the world.
Unit means the properties of the organization where the commuters
work. Table III shows the performance differences between two
Standard errors in parentheses. charge policies and also between single and dual policies.
*, **, *** indicates significance at the 1%, 5% and 10% levels. Generally speaking, if implementing parking and congestion
policies simultaneously, the performance of dual policies is
administration offices, (4) foreign and private enterprise and TABLE III
companies, and (5) other organization. From the survey, we CUMULATED WTP RATIO WITH COMPARISON OF SINGLE OR DUAL
find that part of the transport expense, especially car use POLICIES

expenses can be reimbursed by most of enterprises, Charge rate Cumulated WTP ratio (%)
institutions and offices. 56.94% of the commuters with (yuan/day) Parking charge Congestion charge
car-ownership announce that all or part of car expenses can be 5 36.22 (42.07) 52.16 (56.11)
10 67.10 (72.30) 79.60 (83.03)
reimbursement to their units. To the distribution of expenses, 15 79.81 (84.25) 90.22 (92.11)
30.9% is parking expense, 46.87% is gasoline and 22.23% is 20 92.31 (95.46) 96.00 (97.08)
the other expenses. These facts give evidence that some of car Notes: The ratio of dual policies in parentheses.
use expense will transfer from car users to their units. When
congestion charge policy is carried out, this transfer better than single policy (comparing the figures in and out
mechanism remains working, which can explain why the parentheses in Table III).
properties of units play an important role in the coefficients’ 3) The characteristics of different countries and cities
estimation when ‘Congestion’ policy plays. cannot be neglected when implementing market tools in
practice. From the BLM estimation of two charge policies,
IV. FURTHER ANALYSIS AND CONCLUSIONS some specialties like ‘Unit’ and ‘Personal income’ come out
When increasing cities pay much attention to the marketing when comparing the coefficients of two policies. One of the
tools like parking charge and congestion charge policies to most notable coefficients is the property of units, which
eliminate traffic congestion, it is important to estimate and implies that if the congestion charge would be transferable to
evaluate the performance of these policies when they are the units’ expenses, the effectiveness of congestion charge to
implemented in practice respectively or comprehensively. change mode choice would be weakened. This will be a
Conducting surveys to the road users could get the general significant result that must be estimated in China cities
information of how they would behave and react and what because there are a large number of Government-owned
would be their attitude towards charge policies, which is vehicles (GOVs) running on the road. One of research (Zhang
regarded as a key point to determine the prices reasonably. & Feng, 2008) refers that it is the ratio of GOVs and the ratio
With the aid of the survey to commuters and the BLM of commuters with high time value (or income) that would
estimation, this paper releases some helpful suggestions to the affect the implementation of the congestion charge policy and

509
finally the purpose and expectation of congestion charge
[4] T. Litman, “London congestion pricing: Implication for other cities”,
policy would be baffled. Deeper reform and effective Victoria transport Policy Institute, http://www.vtpi.org/london.pdf.
management of GOVs must be pushed forward to strengthen [5] Municipal Government of Shanghai, Shanghai Metropolitan Transport
and benefit the traffic congestion management. White Paper, Shanghai: Shanghai People Press, 2002.
[6] R. Arnott, A. de Palma, R. Linsey, “A structural model of peak-period
Though obtaining some special and creative insights by
congestion: A traffic bottleneck with elastic demand”, American
comparing different market tool policies using CVM surveys Economic Review, 83, 1993, pp. 161-179.
and BLM estimation, this paper strides some steps towards [7] E.T. Verhoef, P. Nijkamp, P. Rietveld, “Second-best congestion
the pricing and the performance of public policies’ pricing: the case of an untolled alternative”, Journal of Urban
Economics, 40,1996, pp. 279-302.
implementation. Much more angles of public pricing and [8] H. Yang, H.J. Huang, “Principle of marginal-cost pricing: How does it
policy experiment must be introduced into this area. work in a general network?”, Transportation Research A, 32, pp.
45-54.
[9] A.C. Pigou, The Economics of Welfare, Macmillan Company, 1920.
ACKNOWLEDGMENT [10] A.A. Walters, “The theory & measurement of private & social cost of
Very helpful comments by three reviewers are greatly highway congestion”, Econometrica, 29(4), pp. 676-699.
[11] W.S. Vickrey, “Congestion theory and transportation investment”,
acknowledged. All remaining errors are ours alone. American Economic Review, 59, pp. 251-261.
[12] M.Z.F. Li, “The role of speed-flow relationship in congestion pricing
REFERENCES implementation with an application to Singapore”, Transportation
Research B, 2002, 36, pp. 731-754.
[1] T. D. Hau, "Electronic road pricing: Developments in Hong Kong [13] K.W. Axhausen, J.W. Polak, “Choice of parking: Stated preference
1983-89", Journal of Transport Economics and Policy, 24(2), 1990, pp. approach”, Transportation, 1991, 18, pp. 59-81.
203-214. [14] M. Jaccard, W. Haider, K. Washbrook, “Estimating commuter mode
[2] K.A. Small, J.A. Gomez-Ibanez, “Road pricing for congestion choice: A discrete choice analysis of the impact of road pricing and
management: The transition from theory to policy”, in Road Pricing, parking charges”, Transportation, 33,2006, pp. 621-639.
Traffic Congestion and the Environment: Issues of Efficiency and [15] K.E. Train, Discrete Choice methods with Simulation, Cambridge:
Social Feasibility, K.J. Button, E. T. Verhoef, Ed. Edward Elgar Cambridge University Press, 2002.
Publishing Limited, 1998, pp.213-246. [16] W.M. Hanemann, “Valuing the environment through contingent
[3] K.K. Chin, “Road pricing: Singapore’s experience”, paper prepared for valuation”, Journal of Economic Perspectives, 4, 1994, pp. 9-25.
the third seminar of the IMPRINT-EUROPE Thematic Network: [17] Z.A. Zhang, S.W. Feng, “Game analysis of private cars and
“Implementing Reform on Transport Pricing: Constraints and Solutions, government-owned vehicle under road pricing regulation”,
Learning from Best Practice”, Brussels, 2002. Transportation system Engineering & Information Technology, 2008,
8(2), pp. 85-90.

510

You might also like