Scholarly article on topic 'Demand Elasticity to Road Charges in Rome Historical Centre'

Demand Elasticity to Road Charges in Rome Historical Centre Academic research paper on "Civil engineering"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{"demand elasticity" / "limited traffic zone" / "urban road pricing" / "discrete choice models"}

Abstract of research paper on Civil engineering, author of scientific article — Gabriele Giustiniani, Paolo Delle Site, Luca Persia

Abstract An access restriction system (Limited Traffic Zone), in operation from 6.30 a.m. to 6.00 p.m., has been implemented several years ago in the Rome centre to protect the extensive historical and artistic heritage. In order to evaluate the effectiveness of more flexible solutions, combining rationing and pricing policies, several surveys were carried out at night and daytime and demand models were calibrated to simulate possible rationing-pricing schemes. On the basis of such models, direct and cross elasticities of the different options have been analyzed, when a per-trip pricing scheme is applied to cars entering the central area. Elasticities have been evaluated using the “sample enumeration” method. Comparisons have been made between the elasticities of different user groups (systematic vs non-systematic users) and different periods of the day (daytime vs night-time). In daytime, trips made on an occasional basis result to be less elastic to pricing schemes, i.e. respond with behavioural changes relatively less, than trips made systematically. Night-time pricing schemes appear to be more effective in terms of congestion reduction than daytime pricing schemes.

Academic research paper on topic "Demand Elasticity to Road Charges in Rome Historical Centre"

Available online at www.sciencedirect.com

SciVerse ScienceDirect

Procedia - Social and Behavioral Sciences 54 (2012) 1317 - 1329

EWGT 2012

15th meeting of the EURO Working Group on Transportation

Demand elasticity to road charges in Rome historical centre

Gabriele Giustiniani , Paolo Delle Site, Luca Persia

aCTL - Research Centre for Transport and Logistics, University of Rome "La Sapienza", Via Eudossiana 18, Rome 00184, Italy DICEA - Department of Civil Architectural and Environmental Engineering, University of Rome "La Sapienza", Via Eudossiana 18, Rome

00184, Italy

Abstract

An access restriction system (Limited Traffic Zone), in operation from 6.30 a.m. to 6.00 p.m., has been implemented several years ago in the Rome centre to protect the extensive historical and artistic heritage. In order to evaluate the effectiveness of more flexible solutions, combining rationing and pricing policies, several surveys were carried out at night and daytime and demand models were calibrated to simulate possible rationing-pricing schemes. On the basis of such models, direct and cross elasticities of the different options have been analyzed, when a per-trip pricing scheme is applied to cars entering the central area. Elasticities have been evaluated using the "sample enumeration" method. Comparisons have been made between the elasticities of different user groups (systematic vs non-systematic users) and different periods of the day (daytime vs night-time). In daytime, trips made on an occasional basis result to be less elastic to pricing schemes, i.e. respond with behavioural changes relatively less, than trips made systematically. Night-time pricing schemes appear to be more effective in terms of congestion reduction than daytime pricing schemes.

© 2012 Publishedby Elsevier Ltd.Selection and/or peer-reviewunderresponsibilityofthe ProgramCommittee

Keywords: demand elasticity; limited traffic zone; urban road pricing; discrete choice models

* Gabriele Giustiniani. Tel.: +39-0644585131; fax: +39-0644585774. E-mail address: giustiniani@ctl.uniroma1.it

1877-0428 © 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of the Program Committee doi:10.1016/j.sbspro.2012.09.846

1. Introduction (rationing and pricing policies in Rome historical centre)

The implementation of Limited Traffic Zone (LTZ) in Rome started in 1994 in a systematic way, when the transport master plan was implemented and permits to enter LTZ were distributed. These permits were granted by the municipal offices and were given free of charge to residents and other users falling into specific categories. From 1998 some authorised non-residents were required to pay yearly in order to obtain the permit.

In order to enforce traffic limitations and to avoid that drivers could enter the LTZ without permit, in 2001 electronic gates were implemented in the entry points of the areas with traffic restrictions.

Nowadays in Rome there are several LTZs and to enter each of them a permit is needed. The LTZs have a flat-fare pricing scheme. Only authorized categories can have access to the LTZs paying an annual charge. The permit to enter in the LTZ is strictly linked to the plate number of the vehicle and not to the person. Each family that lives inside the LTZ pay for the 1st car about 50 EUR and about 100 EUR for the 2nd, from the 3rd car owned by the family the price is about 550 EUR/year (there are some reductions according to the type of car) .

There are other categories that can be granted a permit and the prices vary between about 100 EUR/year and 550 EUR/year.

In this context, reseach carried out within the European project DIFFERENT ("User reaction and efficient differentiation of charges and tolls"; duration 2006-2008) of the Sixth Framework Programme of Research and Development has estimated the effects of the introduction of a pure pricing (the payment of a price each time a LTZ is entered) on the transport demand. The aim of such research is the analysis of the effects on transport demand due to different price accretions, starting from a minimum reference value (1 EUR).

The procedure has been focused on the aggregated values of demand elasticity for different user categories, characterized according to travel purpose, travel time and destination, on the basis of price variations. The consequent behavioural changes considered involve transport mode change, destination change and time of travel change.

The paper reports on the methodological features of the research, some results from a previous research project on which this work is based, the main results obtained and some comparisons between the elasticity of different user groups (systematic vs non-systematic users) and different periods of the day (daytime vs night-time). Finally, the elasticity values in Rome are compared with those found in other cities.

2. Methodology

In Litman (2008), elasticity is defined as the percentage change in consumption of a good caused by a 1% change in its price (or other characteristics). A negative sign indicates that the effect operates in the opposite direction from the cause (an increase in price causes a reduction in travel). Elasticity can be calculated based on ratios, rather than absolute price values, such as the ratio between transit fares and automobile operating costs, or vehicle costs as a percentage of average income or wages.

There are three different methods commonly found in the transportation literature (Pratt, 2003) for computing elasticity:

• point elasticity;

• arc elasticity;

• shrinkage ratio.

Point elasticity is derived directly from the micro-economics' definition of elasticity and is considered here because disaggregate empirical data are available.

Oum et al. (1992) contend that an important development in transport demand research is the introduction of disaggregate discrete choice models. These models investigate users' travel-choice behavior based on attributes of various transport alternatives, e.g. transport modes, and individual socio-economics characteristics.

Unlike conventional demand models, which assume that consumers make marginal adjustments in response to changes in the environment, discrete choice models assume that consumption is an all-or-nothing decision, e.g. either takes the car or uses public transport.

It is important to note that various demand elasticity measures can be computed from discrete choice models. For example, it is possible to compute an elasticity which measures the percentage change in the probability of a representative individual choosing to travel by bus given a change in transit fare. It is important to note that this is not the same as the regular demand elasticity nor mode-choice elasticity.

Based on empirical experience, Domencich and Mc Fadden (1975) report that the derived regular demand elasticity is likely to be one-half to three-quarters lower than the corresponding representative individual elasticity of choice probability. In order to derive the regular demand elasticity, it is necessary to aggregate across individuals in the population.

Conceptually, a consistent and unbiased estimate of the fraction of population choosing a particular mode is the expected value of the sample probability. In practice, various aggregation procedures are used to approximate the population demand.

A comprehensive review of various aggregation procedures can be found in Ben Akiva and Lerman (1985). Many studies use the sample aggregate as an approximation. The accuracy of this approach clearly depends on the sampling procedure used. Obviously, different procedures will be likely to produce numerically different elasticity estimates. It is therefore important for researchers to state explicitly the aggregation procedure used to derive the aggregate demand and associated elasticities.

In this research, the formulas used to evaluate elasticity are reported below. Formulas (1) and (3) concern the direct elasticity of demand for car with respect to the variation of car price, while formulas (2) and (4) concern the cross elasticity of demand for the other alternatives with respect to the variation of car price.

Disaggregate direct elasticity is referred to car users, and pn(i) is the probability that user n chooses car when a xink car price is experienced.

Disaggregate Direct Elasticity

n index of individual i index of choice alternative k index of attribute E elasticity p probability x attribute

P parameter of the attribute in the utility function

Disaggregate cross elasticity is referred to the other alternatives considered in the model. In formula (2), pn(i) is the probability that user n chooses alternative i, while pn(j) is the probability that user n chooses car when a xjnk car price is experienced. As the models used are multinomial logit, the disaggregate cross elasticity is equal for all alternatives #j.

Disaggregate Cross Elasticity Eji)=-pn(j)• xjnk A

Aggregate elasticity is the elasticity with respect to the expected probability p(i) of alternative i of a population of N individuals (who have different values of one or more attributes), where:

itpn (i)

p(i)=-2=-

Under the hypothesis that the attribute with respect to which the elasticity is evaluated is equal across the population of individuals, we have for the aggregate direct and cross elasticity the following formulas.

Aggregate Direct Elasticity

7~p (i) = dxik = ~

X pn (i)

I pn (i) •Exn(i)

^ =^L_ = J=L_--(3)

lK n=1

Aggregate cross elasticity is given by formula (4) where pn(i) is the probability that user n chooses alternative i when alternative j is priced and xjk is the price level. The aggregate cross elasticity results to be different across alternatives.

Aggregate Cross Elasticity d~p(i) N

Ep(i) _ jk _ n=i__(4)

~p()~ n (4)

— ^ Pn (i)

xjk n_1

The disaggregate and aggregate elasticity series of the different transport alternatives were calculated using the formulas above for a per-trip car price varying between 1 EUR and 6 EUR.

3. Evaluation of the elasticity

3.1. Demand models used

The demand models used for the elasticity evaluation were developed within the activities on Rome historical centre of the European project PROGRESS ("Pricing road use for greater responsibility, efficiency and sustainability in cities"; duration 2000-2004) of the Fifth Framework Programme of Research and Development.

The demand models are logit and were calibrated using Revealed Preferences (RP) and Stated Preferences (SP) surveys of different user categories. Survey activities were carried out between October 2001 and November 2002.The demand models were part of a larger system of simulation models, including supply models and assignment models, which were developed in PROGRESS to assess the impacts on mobility of different pricing

schemes in the centre of Rome.

Despite the relatively limited area involved in the pricing intervention, the high attractiveness of the LTZ has induced to consider a much larger area of study (the whole municipality of Rome and beyond) in the traffic simulations. A graph with about 16.900 directed arcs and about 5.900 nodes has been considered as base network and about 700 Public Transport (PT) lines (including buses, tramways, metro, and the regional railway) have been loaded on it.

Starting from this base network, three modal networks have been constructed: the first one for private cars, characterized by standard BPR (Bureau of Public Roads, 1964) arc cost functions retrieved from an existing graph, the second one for motorbikes, the third one for PT, where the base network acts as pedestrian network.

Two charging schemes, differentiated by time of day, were considered for the estimation of the demand models: a daytime scheme ("Daytime Scheme") and a night scheme. The night scheme is applicable to the summer period. For this reason, it will be referred to hereafter as "Summer Scheme". Beyond the introduction of road charges, also the improvement of PT service (quantified in terms of trip time) has been considered.

In the Daytime Scheme the simulation activities considered two choice levels: mode choice and route choice. In the route choice model the generalized cost of each alternative is evaluated by summing up the contribution of two terms: monetary cost and value of time. Comfort has been taken into account by multiplying the value of time by class-specific factors depending on the trip phase represented (driving a car, driving a motorbike, travelling on a bus, waiting at a PT stop, walking, and so on). A hyper path route choice model has been used on the PT network.

The mode choice model has considered two classes of users: systematic users and non-systematic users. The latter represents the group of those travelling occasionally. Moreover, as the main purpose of the simulations performed is to evaluate the effects of pricing on the travel choices of users directed to the LTZ, a specific demand model for this class has been estimated. Four different modes are included in the user choice sets: car, moped, PT and park and ride (P&R).

The RP data have been used to calibrate the whole utility model, except for the charge coefficient. The latter one was obtained from the SP data, where the results gathered from the RP have been used for representing the utility of the current state (the SP were based on modifications of the current state obtained by imposing charges at different rates for entering the LTZ).

In the Daytime Scheme the simulation activities focused on the results of the application of different road pricing schemes to the defined area in the current operational time of the LTZ (from 6:30 am to 6:00 pm).

The simulation of a scenario where the complete substitution of the current access restriction scheme with a "pure" road pricing scheme is assumed showed that, in order to have a modal split similar to the current one (based on access restriction), a very high charge should be applied (more than 32 EUR per trip).

In the second wave of simulation activities, the effects of a road pricing scheme application in the night period (from 6:00 pm to 11:00 pm) have been assessed.

In the Summer Scheme, some scenarios focused on the isolated effects of the road pricing scheme introduction, while in other scenarios the combined effects of both road pricing and complementary measures (i.e. PT supply increase) are investigated. In all the simulated scenarios, except for the current scenario, a per-trip charging structure is assumed. Levels of prices are common for all car users. Charges are not applied to residents, taxis and public utility vehicles.

The unavailability of reliable origin-destination matrices did not allow to perform a complete simulation, taking into account destination, mode and route choice models. Particularly, the route choice model, and, thus, the effects of the congestion, has not been considered. This assumption, although reducing the precision of the model, has been considered acceptable, since in the evening hours the congestion level (for both private and PT network) is lower than in the morning/afternoon hours and, thus, lower is its influence on the higher level user choices (destination of the trip and used transport mode). Obviously, the simulation results are obtained in terms of aggregated data (e.g. overall modal split) rather than in terms of flows on each link of the networks. Only the

"change of route" for a particular class of users has been considered among the user choice set.

Multinomial Logit models, calibrated on RP-SP data, have been used to assess the impacts on user behaviour of the introduction of road pricing in the Summer Scheme.

In the Summer Scheme, four different choice models have been calibrated according to the class of users:

• users with destination inside the LTZ for work purposes (briefly "work trips");

• users with destination inside the LTZ for recreational purposes (briefly "recreational trips");

• users with destination inside the LTZ for shopping purposes (briefly "shopping trips");

• general car drivers currently crossing the LTZ (briefly "crossing trips").

In the user choice sets of the Summer Scheme, not only the different modes (car, moped, PT) were included, but also, according to the different classes of users, change of destination, postponement of the trip after 11.00 p.m. (i.e. when the charge would be no longer applicable), and change of route (i.e. avoiding the charged area).

Table 1 shows the segmentation of demand and the alternatives considered in each model. Table 2 shows the attributes included in the utility functions. Table 3 and Table 4 the results of the estimation.

Table 1. Segmentation of demand and alternatives considered in each model

Car PT Moped P&R Change of destination Delay in departure Change of route

Daytime scheme

Systematic generic user X X X X

Non-systematic generic user X X X X

Systematic LTZ user X X X X

Non-systematic LTZ user X X X X

Summer scheme

Work trips X X X X

Shopping trips X X X X

Recreational trips X X X X X

Crossing trips X X X X

Table 2. Attributes considered in the utility functions

Daytime scheme travel time (min) origin or destination in LTZ (0/1)

sum of distance to the closest metro stop form origin and destination (km) rush hour (0/1)

monthly frequency (times/month)

distance from origin to destination (km)

duration of stay (h)

female (0/1)

employee (0/1)

student (0/1)

old (0/1)

fine (EUR) charge (EUR) Summer Scheme male (0/1) class of age (1-4)

class of duration of stay (h) inside the LTZ (1-4)

monthly frequency of the trip (times/month)

travel time by car (min)

travel time by moped (min)

travel time by public transport (min)

charge (EUR)

Table 3. Estimation results: statistics

Daytime scheme

systematic non-systematic systematic LTZ non-systematic LTZ

Number of observations 809 1196 352 761

Initial Likelihood -897.2929 -1164.7162 -398.536 -748.2523

Final value of Likelihood -621.7591 -611.0232 -185.127 -297.742

"Rho-Squared" w.r.t. Zero 0.3071 0.4754 0.5355 0.6021

"Rho-Squared" w.r.t. Constants 0.3024 0.4257 0.4469 0.4836

Summer scheme

Work Recreational Shopping Crossing

Number of observations 608 1527 450 430

Initial Likelihood -765.3544 -2232.2718 -562.2506 -596.1066

Final value of Likelihood -549.5873 -1682.4186 -367.0592 -274.8396

"Rho-Squared" w.r.t. Zero 0.2819 0.2463 0.3472 0.5389

"Rho-Squared" w.r.t. Constants 0.2373 0.1733 0.2749 0.1609

Table 4. Estimation results: coefficients

Daytime scheme

systematic non-systematic systematic LTZ non-systematic LTZ

Coefficient Estimate T Ratio Estimate T Ratio Estimate T Ratio Estimate T Ratio

ASA_M -4.628 -6.600 -9.645 -8.600 0.590 0.400 -3.958 -3.500

ASA_TC -2.349 -3.600 -7.643 -9.800 5.517 3.100 -3.631 -3.900

ASA_PR -3.106 -4.400 -7.230 -8.900 2.692 1.400 -4.241 -4.300

bZTL_M 2.032 6.400 2.013 3.600

bZTL_PUB 0.914 3.300 0.888 4.500

bMETRO_M 0.058 0.500 -0.321 -1.400 -0.182 -0.400 -0.050 -0.200

bMETRO TC 0.182 1.800 -0.176 -2.400 -0.566 -1.200 -0.014 -0.100

bMETRO_PR -0.097 -0.900 -0.472 -5.100 -0.613 -1.300 -0.159 -0.600

bRUSH_M -0.483 -1.400 0.625 1.400 -0.729 -0.800 0.635 1.200

bRUSH_TC -0.500 -1.500 1.471 7.100 0.904 1.000 1.981 4.500

bRUSH_PR -0.541 -1.400 0.631 2.300 0.057 0.100 1.126 2.200

bFREQ_M 0.031 1.500 0.079 3.200 0.071 1.000 0.056 2.000

bFREQ_TC 0.065 3.100 0.125 9.200 0.248 3.100 0.183 6.700

bFREQ_PR 0.111 4.500 0.122 7.300 0.293 3.400 0.109 3.600

bDIST_M -0.079 -2.300 -0.007 -0.100 -0.101 -1.100 -0.033 -0.400

bDIST_TC 0.027 1.000 0.090 3.800 0.115 1.200 0.131 2.300

bDIST_PR 0.132 4.500 0.185 6.300 0.186 1.900 0.227 3.900

bSTAY_M -0.067 -1.300 -0.052 -0.400 -0.231 -1.400 -0.220 -1.400

bSTAY_TC -0.310 -5.500 -0.093 -1.300 -1.426 -5.900 -0.520 -3.500

bSTAY_PR -0.553 -6.800 -0.048 -0.500 -1.191 -4.700 -0.157 -1.000

bFEMALE_M -0.030 -0.100 -0.263 -0.600 0.409 0.500 0.237 0.500

bFEMALE_TC 0.642 2.600 0.371 1.800 0.579 0.700 1.380 3.400

bFEMALE_PR 0.142 0.500 -0.416 -1.600 0.447 0.500 0.477 1.100

bSTUD_M 1.442 4.200 0.847 2.000 2.783 3.500 0.913 2.000

bEMP_PUB -1.192 -4.300 -0.323 -1.700 -2.558 -3.500 -0.674 -1.800

bOLD_PUB -0.206 -0.900 0.805 4.100 -1.921 -3.200 0.969 2.500

bTIME_A -0.035 -6.800 -0.029 -6.600 -0.006 -0.300 -0.015 -2.200

bTIME_M -0.014 -1.400 -0.028 -1.200 0.036 1.400 -0.037 -1.500

bTIME_TC -0.043 -7.800 -0.041 -8.400 -0.049 -5.700 -0.057 -7.700

bTIME_PR -0.028 -4.500 -0.033 -5.000 -0.026 -2.400 -0.032 -3.500

bFINE -0.063 -9.100 -0.119 -9.100 -0.082 -4.700 -0.102 -8.200

bCHARGE -0.085 -7.000 -0.108 -9.100 -0.095 -9.100 -0.104 -10.700

Summer scheme

Work Recreational Shopping Crossing

Coefficient Estimate T Ratio Estimate T Ratio Estimate T Ratio Estimate T Ratio

ASA tp 1.3690 2.5 0.8900 2.7 0.6769 0.9 - -

ASA mop -0.2238 -0.4 1.2940 3.4 1.1630 1.5 - -

ASA perc - - - - - - -1.9190 -3.9

ASA dest -0.4500 -0.7 -0.2347 -0.8 -3.0730 -4.1 -5.7630 -4.6

ASA rit - - -2.0600 -3.3 - - -3.8630 -2.7

ASA tpRP -0.6402 -1.1 -2.5850 -6.3 -1.0650 -1.4 - -

ASA tmop -0.7137 -1.1 -1.2560 -2.8 1.1580 1.5 - -

bsex tp 0.7625 -2.5 -0.3935 -2.4 -03579 -1.0 - -

bsex mop 0.2287 -0.6 -0.1696 -0.9 0.4018 1.1 - -

bsex dest -1.4240 -3.9 -0.7459 -0.4 0.6601 1.4 0.2842 0.4

bsex rit - - 0.7808 2.1 - - 0.5521 0.5

bsex perc - - - - - - 0.3700 0.1

bage tp -0.2804 -1.6 -0.2913 -3.1 -0.5009 -1.9 - -

bage mop -0.5830 -0.3 -0.7865 -6.1 -0.9369 -3.2 - -

bage dest -0.5348 0.2 -0.1386 -1.4 0.5278 1.8 0.5206 1.4

bage rit - - -0.5375 -2.6 - - -0.8411 -1.7

bage perc - - - - - - -0.1540 -1.0

bstay tp -0.6349 0.8 0.1166 1.2 0.3834 1.5 - -

bstay mop 0.3364 3.9 0.2591 2.4 0.1038 0.4 - -

bstay dest 0.0000 0.0 0.0000 0.0 0.0000 0.0 - -

bstay rit - - 0.2616 1.5 - - - -

bfreq tp 0.0138 0.1 -0.0227 0.2 0.7314 2.2 - -

bfreq mop -0.0208 -0.2 -0.0469 -0.4 0.2665 0.7 - -

bfreq dest 0.0000 0.0 0.0000 0.0 0.0000 0.0 0.0000 0.0

bfreq rit - - -0.0902 -0.4 - - 0.1743 0.5

bfreq perc - - - - - - 0.0000 0.0

bt-car -0.0345 -4.6 -0.0656 -2.6 -0.0696 -1.9 - -

bt-tp -0.0289 -7.1 -0.0160 -9.6 -0.0251 -7.1 0.0080 0.7

bt-mop -0.0334 -7.9 -0.0283 -10.8 -0.0264 -6.4 - -

Charge -0.1030 -2.4 -0.4831 -2.9 -0.3997 -4.8 -1.1470 -7.7

The two considered scenarios (the "daytime" one and the "night" one) are completely different for several reasons:

• in the first one, only selected groups of car drivers (residents, authorised car drivers and public utility vehicles) are allowed to have access to the area, while in the second one the access is completely free;

• as a consequence of the previous item, in the night hours a quite large set of car drivers just cross the area, without having their destination inside it;

• in the daytime hours systematic trips are prevailing, while in the evening period most of the trips are for recreational or for shopping purposes.

3.2. Results

In the Daytime Scheme only two different elasticity series have been calculated according to the two different models calibrated that are:

• LTZ systematic users;

• LTZ non systematic users.

For the other two models calibrated it was not possible to calculate the elasticity because car price was not a variable of the models.

Figure 1 shows the elasticity variations for each alternative considered in the Daytime Scheme models when a car price varying from EUR 1 to EUR 6 is experienced by the car trips entering in the LTZ.

0,12 0,1 0,08 0,06 0,04 0,02 0

-0,02 -0,04 -0,06 -0,08

0 sys - car 3 sys - moped

□ sys - PT H sys - P &R

□ non sys - car

a non sys -moped B non sys - PT [D non sys - P &R

Figure 1. Daytime Scheme: elasticities of demand for the transport modes vs car price level

An overview on the results shows that both LTZ systematic and non systematic users have rigid behaviour when an increase of the car price is experienced. Anyhow, the results show some minor differences between LTZ systematic and non systematic users.

A comparison shows that LTZ systematic car users have elasticity varying from -0.01 to -0.07 for a car price varying from EUR 1 to EUR 6. In contrast, LTZ non systematic car users have elasticity varying from 0 to -0.03 for the same car price variation. This comparison shows that LTZ non systematic car users have a slightly more rigid behaviour than LTZ systematic users. Probably the reasons are that LTZ systematic car users can plan their trip and they cannot afford the car price daily.

Moped users (systematic and non systematic) have behaviours almost inelastic for each car price, and car drivers do not shift to moped probably because who owns a moped already use it.

Concerning the Summer Scheme, Figure 2 shows the variation of the elasticities of demand for car and change of destination when a car price varying from EUR 1 to EUR 6 is experienced for recreational, work and shopping trips entering the LTZ. The PT, moped and delay in departure elasticities are not shown because they are always, in absolute terms, lower than the car and change of destination elasticities and don't show appreciable variations with car price.

Shopping trips show a strong increase of elasticity of demand for car, in absolute terms: elasticity is about -2 for a 6 EUR car price. Change of destination elasticity also shows a quite significant increase when a car price increase is experienced.

Work trips, with the increase of car price show small modifications, in absolute terms, of car and change of destination elasticities, and elasticities are always, in absolute terms, much lower than one both for car and change of destination alternatives.

Recreational trip elasticities do not show appreciable variations when a car price increase is experienced and they are very low in absolute terms.

An overview on the Summer Scheme results shows that there are differences among the different models. Concerning car these differences are quite significant. When car price varies from EUR 1 to EUR 6, work trip elasticities vary from -0.07 to -0.47, shopping trip elasticities vary from -0.22 to -1.99 and recreational trip elasticities vary from -0.03 to -0.19. It means that users moving for recreational purposes have more rigid behaviour than the users moving for work and shopping purposes.

0,50 0,00 -0,50 -1,00 -1,50 -2,00 -2,50

Tl % 1 , 11 " 1 f 1 ji

□ car- recreationaI

Hear- work

El ca r - s hopping

S change ofdestination-recreational

13 change ofdestination-work

H change ofdestination-s hopping

Figure 2. Summer Scheme: elasticities of demand for car and for destination change vs car price level.

Figure 3 shows the elasticity values for crossing trips.

2,00 1,00 0,00 -1,00 -2,00 -3,00 -4,00 -5,00 -6,00 -7,00 -8,00

0 cha nge of des ti nation DD delay in depa rture IS cha nge of route

Figure 3. Summer Scheme - crossing trips: elasticities of demand for car and for other travel choices vs car price level.

For crossing trips the model shows very high car elasticity, in absolute terms, which varies from -0.44 to -6.85 when car price varies from EUR 1 to EUR 6. In this case it is necessary to evaluate such result in relation with the different dimensions of choice. In fact the car users do not leave the car but they prefer changing destination, delaying the departure or changing the route. The cross elasticities increase for a 1 EUR or 2 EUR price, then decrease for higher car prices probably because the increase of car price does not determine a proportional increase of alternatives modal shares.

3.3. Comparison of results between Daytime Scheme and Summer Scheme

A comparison has been made between the Daytime Scheme LTZ systematic user model results and Summer Scheme work trips model results. The LTZ non systematic user model results have been compared with shopping and recreational trip model results of the Summer Scheme.

LTZ systematic users are more rigid in terms of elasticity of demand for car than users of the work trip model: when car price increases from EUR 1 to EUR 6, the LTZ systematic user elasticity varies from -0.01 to -0.07, while the work trip elasticity varies from -0.07 to -0.47. The PT and moped elasticities of the LTZ systematic users and of the work trip models don't show many differences, being the elasticity values always lower than 0.10. LTZ non systematic user and recreational trip models show very similar results: in both elasticities are always lower than 0.09 and there are no significant behavioural differences.

r ir- mi

I 0 B

shopping

recreationa I

i systematic LTZ users

Figure 4. Comparison of elasticities of demand for car between Daytime Scheme and Summer Scheme.

Concerning crossing trips in the Summer Scheme, the elasticity of demand for car varies from -0.44 to -6.85 when the car price increases from EUR 1 to EUR 6.

3.4. International comparisons

The DIFFERENT project collected evidence on transport demand elasticities according to the different prices in different foreign contexts. In London, Trondheim and Singapore the ex-post evaluation of the elasticity of demand for car linked with the introduction of a price in the daytime hours has been carried out. The existing Trondheim price is closer to the minimum of the range considered for Rome, 1 EUR, while the London price is closer to the maximum of the Rome range, 6 EUR. The comparison between the elasticity values estimated for Rome and those resulting from the evaluations in London and Trondheim shows that the demand for car in the Daytime Scheme in Rome can be considered as not elastic. With a 1 EUR price, the systematic user elasticity in Rome is about -0.01 (it is 0 for non-systematic users), whereas in Trondheim (where the price is 0.97 EUR) the elasticity is about -0.32. In the same way, with a 6 EUR price the systematic user elasticity in Rome is about -0.07 (it is -0.03 for non-systematic users), whereas in London (where the price is 6.35 EUR) the elasticity is about -1.30. In Singapore the transport demand elasticity has been evaluated by varying the price between 0.25 EUR and 1.50 EUR. The resulting elasticity values (0 for a 0.25 EUR price and -0.42 for a 1.50 EUR price) are

indicative of a slightly more elastic car user demand than in Rome (both for LTZ systematic and non-systematic users). Table 5 shows the summary of the comparisons made.

Table 5. International comparisons (Rome: Daytime Scheme)

City Price level [EUR] Elasticity of demand for car Ex-ante Ex-post

Rome 1.00-H5.00 (Sys) -0.01^0.07 (Non sys) -0^0.03 -

London 6.35 - -1.03

Trondheim 0.97 - -0.32

Singapore 0.25^1.50 - 0^-0.42

4. Conclusions and further analysis developments

For the Daytime Scheme results show that both systematic and non-systematic LTZ users have very rigid behaviour when a car price increase is experienced and that the non-systematic LTZ car users have a more rigid behaviour than the systematic ones probably because LTZ systematic car users can plan their trip and they cannot afford the car price daily. For the Summer Scheme results show that, when a car price increase is experienced, shopping trip car elasticity is the most affected, followed by work trip car elasticity and recreational trip car elasticity. On the other hand, PT and moped cross elasticities are not so affected as the change of destination elasticity probably because the change of destination alternative shows a larger relative increase in terms of share. Crossing trips in the Summer Scheme are strongly affected when a car price increase is experienced and car elasticity is very high in absolute terms. For crossing trips it is necessary to evaluate such result in relation with the different dimensions of choice. In fact, the car users do not leave the car but they prefer changing destination, delaying the departure or changing the route.

Overall, it appears that a per-trip car pricing scheme during the daytime would not be as effective in reducing car use as the current mix between permits and flat-fare pricing scheme is. In contrast, for the night-time schemes with different characteristics of travellers a charge per trip would probably be somehow effective in reducing congestion. The rigid behaviour of the car drivers during the daytime when a car price increase is experienced needs probably further analysis. One possible hypothesis could be that the Rome PT system is not considered a possible alternative to private vehicles for many LTZ systematic and non systematic users. To investigate this hypothesis, it would be useful to organize focus groups with stakeholders such as LTZ systematic and non systematic users and to analyze in detail the results from other cities where road pricing has been implemented.

References

Ben Akiva M. and Lerman S.R. (1985) Discrete Choice Analysis: Theory and Application to Travel Demand, The MIT Press. Bureau of Public Roads (1964) Traffic Assignment Manual. U.S. Dept. of Commerce, Urban Planning Division, Washington D.C.. Domencich T.A. and McFadden D. (1975) Urban Travel Demand: A Behavioral Analysis. North Holland Publishing Company - Amsterdam

Oxford, American Elsevier Publishing Company, Inc. New York. Litman T. (2008) Transportation Elasticities: How Prices and Other Factors Affect Travel Behavior. Victoria Transport Policy Institute. Oum T.H., Waters II W.G. and Yong J.-S. (1992) Concepts of Price Elasticities of Transport Demand and Recent Empirical Estimates: An

Interpretative Survey. Journal of Transport Economics and Policy 26(2), 139-154. Pratt R.H. (2003) Traveler Response to Transportation System Changes: An Interim Introduction to the Handbook, Appendix A. In: Research Results Digest: TCRP Report 95, Chapter 14.Transportation Research Board of National Academies.