Congestion management using road pricing: Would it be efficient?

0 downloads 0 Views 77KB Size Report
inverse demand curves between the actual and efficient levels of traffic. Potential .... DWL and MCC in cents per pcu-mile increase as the cordon is reduced and,.
Congestion management using road pricing: Would it be efficient? Georgina Santos [email protected] Department of Applied Economics Cambridge CB3 9DE, UK August 2000

Abstract A traffic simulation and assignment model is used to compute the social costs of congestion in five English towns, together with potential road charges. The social costs of congestion are measured by the deadweight loss and computed as the area between the marginal social cost and inverse demand curves between the actual and efficient levels of traffic. Potential road charges are computed as the difference between average private and marginal social costs at the efficient level of traffic, defined as the level at which marginal benefit, given by the inverse demand curve, and marginal cost, are equal. The exercise is done for different demand curves and elasticity values and for different areas within each town.

Key words Road pricing, traffic congestion, congestion charges, congestion externality, congestion costs.

Introduction Traffic demand management has a priority place on the current agenda of governments in many countries throughout the world and in particular in the UK, where a Transport Bill (House of Commons, 1999), issued by the UK Government in December 1999, authorizes local authorities to introduce road user charges and workplace parking levies to help tackle congestion in towns and cities. Although it has not received Royal Assent yet, the Bill is currently undergoing the different legislative stages and will probably be passed if not this parliamentary session, next (beginning November 2000). The Bill specifies that charges should only be introduced when they appear ‘desirable for the purpose of directly or indirectly facilitating the achievement of policies in the charging authority’s local transport plan’ (Part III, Chapter I, Clause 140.2) but it does not state how the magnitude of these charges should be determined. Newbery and Santos (1999) sustain that these charges appear to be additional to the already excessive road taxes, and that there seems to be no plan of regulating them to protect against further exploitation. Road pricing has a strong theoretical basis but congestion costs must be quantified before it can be sensibly implemented. The SATURN (Simulation and Assignment of Traffic to Urban Road Networks) model is used to compute the social costs of congestion in five English towns, together with potential road charges. The exercise is done for different demand curves and elasticity values and for different areas within each town.

1

Model used

Average and Marginal Costs (cents/ pcu-mile)

The efficient charge should equal the external cost imposed on other drivers, or difference between marginal social cost (MSC) and average private cost (APC). This is standard economic theory and can be summarised on Figure 1. Qa is the actual level of traffic, which is inefficient because the marginal cost QaH is greater than the marginal benefit, QaC. The externality or marginal congestion cost (MCC) in this case is HC. At this point, drivers are not paying for the full costs of their trips. By introducing a charge equal to BD, the demand for trips will be reduced to Qe and marginal cost will be equal to marginal benefit. The social costs of congestion are the deadweight loss (DWL) or inefficiency due to incorrect pricing of scarce road resources. This can be measured by the difference between the marginal social cost and the inverse demand curve between the actual and efficient levels of traffic. Graphically, it is the triangle BHC on Figure 1.

MSC Demand H

E

B APC C D

A

Qe

Qa

Traffic load on the network (pcu/h)

Figure 1: Average private cost, marginal social cost, marginal congestion cost, efficient charge and deadweight loss

The SATURN suite (Simulation and Assignment of Traffic to Urban Road Networks), developed at the Institute for Transport Studies at Leeds University (Van Vliet and Hall, 1997), was run for different levels of traffic in five English towns for the morning peak. Given an origin-destination (O-D) matrix and a specification of the road network (characteristics of links and junctions) the model attempts to find the equilibrium traffic flows which are characterized by the situation in which no road user can lower his trip cost (vehicle cost plus cost of time taken) by altering his journey. Passenger car units (pcu), which represent a weighted average of five vehicle categories (cars, light good vehicles, two types of other good vehicles and public service vehicles) were used instead of vehicles. Thus, the units in the trip matrix are pcu’s, not vehicles. A matrix of costs to go from origin zone i to destination zone j was produced for each town using one of the programs within the SATURN suite. The marginal social cost

2

Average and Marginal Costs (cents/pcu-mile)

(MSC) and average private cost (APC) were then computed with Matlab. Thus we obtained the values for Qa, C and H. Then we assumed an inverse demand function and a point elasticity for the actual level of traffic and computed the values of Qe, B and D, and the efficient charge. We also computed DWL as triangle BHC (expressed in cents/milehour) in all cases and converted it in order to have it expressed it in cents per pcu-mile and in $ million per year. All values depend on the demand function and elasticity values assumed. Three different inverse demand functions were considered: linear, elastic exponential and constant elasticity. Figure 2 shows the three curves.

Constant Elasticity

Exponential

MSC

Linear

APC

Traffic load on the network (pcu/h)

Figure 2: Different demand functions

The demand functions are presented in Table 1 together with their elasticities. Five point elasticities at the actual level of traffic were considered in this study: 0.2, 0.3, 0.4, 0.5, 0.6 and 0.7. Table 1: Different demand functions and their elasticities Inverse demand function

Demand function Constant Elasticity

P Q( P ) = Q0 *   P0

Elastic Exponential

Q( P ) = Q0 * e

Linear

Q( P ) =

  

p

 P  ρ *  − 1   P0 

Q P( Q ) = P0 *   Q0

  

1/ p

1 Q  P( Q ) = P0 *  ln + 1  Q ρ 0  

P( Q ) − a b

P( Q ) = a + bQ

3

Elasticity

η=ρ ρ* P P0 P η= bQ

η=

The areas covered Most local authorities provided us with the SATURN model calibrated for the town including the motorways and trunk roads that surround it. Three cordons were used in the simulations. The first and second ones followed approximately the contour of the motorways and trunk roads surrounding the town, including and excluding them respectively. Simulations were also performed for the very central area of each town. This is the most congested area and in most cases we spoke to the local authority to make sure we were indeed simulating traffic in what they would define as the central area.1 When a cordon is placed in SATURN, the model considers that trips begin at the edge of the cordon. Trips beginning and finishing outside are included but they are assumed to begin and end at the edge. The basic principle is that all trips that cross the cordon and enter the area under study are taken into account regardless of where they start or finish.

Results Tables 2 to 6 show the results obtained for Cambridge, York, Norwich, Lincoln and Exeter. Values of DWL (in cents per pcu-mile) and annual DWL (in $ million) are presented for six elasticity values for each of the three demand functions. MCC in cents per pcu-mile are also given for the different cordons.2 As it can be seen from the tables, the following conclusions can be drawn: 1. DWL and MCC in cents per pcu-mile increase as the cordon is reduced and, consequently, encircles more congested areas. 2. Annual DWL may decrease or increase as the encircled area decreases. Since annual DWL depends on the product of DWL in cents per pcu-mile and total vehicle-miles traveled (VMT), it may actually increase if DWL grows more rapidly than VMT decreases. 3. For a given cordon, DWL and annual DWL increase as the absolute value of the elasticity increases.

1

Most local authorities define an inner ring road and the area inside was considered to be the central area of the town or city center. 2 Original calculations used British units (pence/pcu-km and pounds million). For the sake of this paper they were converted to American units at the rate 1 pence = $ 0.16 and 1 km = 0.621 miles.

4

Table 2: Results for Cambridge Point elasticity at Annual DWL for the morning DWL in cents per pcu-mile peak in $ million the original level of traffic Area under study Area under study Includes Excludes Central Includes Excludes Central motorways motorways area motorways motorways area Constant 0.2 4.48 4.48 2.24 10.82 25.86 46.11 Elasticity 0.3 5.44 5.44 2.56 12.83 30.14 52.55 Demand 0.4 6.08 6.08 2.72 14.17 32.97 56.41 0.5 6.45 6.37 2.88 15.20 35.03 58.99 0.6 6.72 6.56 2.88 15.97 36.32 60.79 0.7 7.04 6.88 3.04 16.74 37.35 62.34 Elastic 0.2 4.96 5.12 2.40 11.59 28.08 49.72 Exponential 0.3 5.76 5.76 2.72 13.65 31.94 54.87 Demand 0.4 6.24 6.24 2.88 14.94 34.26 58.22 0.5 6.72 6.56 2.88 15.71 35.81 60.28 0.6 6.88 6.72 3.04 16.49 37.09 61.82 0.7 7.20 6.88 3.04 17.00 38.12 63.11 Linear 0.2 4.96 5.28 2.40 11.85 28.59 50.23 Demand 0.3 5.76 5.92 2.72 13.91 32.46 55.64 0.4 6.37 6.32 2.88 15.20 34.78 58.73 0.5 6.72 6.56 2.88 15.97 36.32 60.79 0.6 7.04 6.88 3.04 16.74 37.61 62.34 0.7 7.20 7.04 3.04 17.26 38.38 63.63 Average 6.2 6.2 2.8 14.8 34.0 57.6 Standard 0.8 0.7 0.2 2.0 3.7 5.1 deviation MCC in cents per pcu-mile 198.1 426.3 717.1 Source: Own calculations, see text

5

Table 3: Results for York Point elasticity at Annual DWL for the morning DWL in cents per pcu-mile peak in $ million the original level of traffic Area under study Area under study Includes Excludes Central Includes Excludes Central motorways motorways area motorways motorways area Constant 0.2 1.76 1.44 0.96 4.89 9.27 42.76 elasticity 0.3 2.08 1.76 1.12 5.92 11.33 48.43 demand 0.4 2.40 1.92 1.28 6.96 12.62 51.78 0.5 2.72 2.08 1.28 7.47 13.40 54.10 0.6 2.88 2.24 1.28 7.99 14.43 55.64 0.7 3.04 2.40 1.28 8.50 14.94 56.93 Elastic 0.2 1.92 1.58 1.12 5.26 10.05 45.60 exponential 0.3 2.24 1.92 1.28 6.44 11.85 50.23 demand 0.4 2.56 2.08 1.28 7.21 13.14 53.07 0.5 2.72 2.24 1.28 7.78 14.17 55.13 0.6 2.88 2.40 1.44 8.24 14.68 56.41 0.7 3.04 2.40 1.44 8.76 15.20 57.44 Linear 0.2 1.92 1.62 1.12 5.41 10.25 46.11 demand 0.3 2.24 1.92 1.28 6.44 12.11 50.75 0.4 2.56 2.08 1.28 7.21 13.40 53.58 0.5 2.72 2.24 1.28 7.99 14.17 55.64 0.6 2.88 2.40 1.44 8.50 14.94 56.93 0.7 3.04 2.40 1.44 8.76 15.47 57.96 Average 2.5 2.1 1.3 7.2 13.1 52.7 Standard 0.4 0.3 0.1 1.2 1.9 4.5 deviation MCC in cents per pcu-mile 111.7 198.4 751.4 Source: Own calculations, see text

6

Table 4: Results for Norwich Point elasticity at Annual DWL for the morning DWL in cents per pcu-mile peak in $ million the original level of traffic Area under study Area under study Includes Excludes Central Includes Excludes Central motorways motorways area motorways motorways area Constant 0.2 0.8 1.12 0.48 1.55 11.08 31.43 elasticity 0.3 1.12 1.28 0.64 1.80 13.40 35.94 demand 0.4 1.28 1.44 0.64 2.32 14.94 38.61 0.5 1.52 1.60 0.64 2.59 16.23 40.44 0.6 1.76 1.76 0.64 2.83 17.00 41.73 0.7 1.76 1.76 0.64 3.09 17.77 42.76 Elastic 0.2 0.96 1.28 0.64 1.55 12.11 33.59 exponential 0.3 1.12 1.44 0.64 2.06 14.17 37.35 demand 0.4 1.44 1.60 0.64 2.32 15.46 39.67 0.5 1.58 1.60 0.64 2.68 16.49 41.16 0.6 1.76 1.76 0.64 2.83 17.26 42.25 0.7 1.92 1.76 0.80 3.09 18.03 43.28 Linear 0.2 0.96 1.28 0.64 1.55 12.11 34.00 demand 0.3 1.28 1.44 0.64 2.06 14.43 37.87 0.4 1.44 1.60 0.64 2.32 15.71 39.93 0.5 1.62 1.76 0.64 2.83 16.74 41.47 0.6 1.76 1.76 0.64 3.09 17.52 42.76 0.7 1.92 1.76 0.80 3.09 18.29 43.53 Average 1.4 1.6 0.6 2.4 15.5 39.3 Standard 0.3 0.2 0.1 0.6 2.2 3.6 deviation MCC in cents per pcu-mile 54.1 243.7 581.2 Source: Own calculations, see text

7

Table 5: Results for Lincoln Point elasticity at Annual DWL for the morning DWL in cents per pcu-mile peak in $ million the original level of traffic Area under study Area under study Includes Excludes Central Includes Excludes Central motorways motorways area motorways motorways area Constant 0.2 2.24 1.76 1.76 10.05 11.08 46.63 elasticity 0.3 2.72 2.24 2.08 12.36 13.40 53.58 demand 0.4 3.04 2.40 2.24 13.65 14.94 57.96 0.5 3.36 2.56 2.24 14.68 15.97 60.79 0.6 3.52 2.72 2.40 15.53 17.00 62.85 0.7 3.68 2.88 2.40 16.23 17.52 64.35 Elastic 0.2 2.56 1.92 1.92 11.08 12.11 50.49 exponential 0.3 2.88 2.40 2.08 12.96 14.17 56.16 demand 0.4 3.22 2.56 2.24 14.17 15.53 59.76 0.5 3.36 2.72 2.40 15.20 16.49 62.08 0.6 3.52 2.88 2.40 15.97 17.26 63.88 0.7 3.68 2.88 2.40 16.49 17.90 65.17 Linear 0.2 2.56 2.08 1.92 11.33 12.36 51.26 demand 0.3 3.04 2.40 2.08 13.14 14.43 56.93 0.4 3.26 2.56 2.24 14.43 15.71 60.54 0.5 3.49 2.72 2.40 15.46 16.74 62.85 0.6 3.68 2.88 2.40 16.23 17.52 64.66 0.7 3.84 3.04 2.40 16.74 18.29 65.95 Average 3.2 2.5 2.2 14.2 15.5 59.2 Standard 0.5 0.4 0.2 2.0 2.1 5.7 deviation MCC in cents per pcu-mile 197.8 218.4 718.8 Source: Own calculations, see text

8

Table 6: Results for Exeter Point elasticity at Annual DWL for the morning DWL in cents per pcu-mile peak in $ million the original level of traffic Area under study Area under study Includes Excludes Central Includes Excludes Central motorways motorways area motorways motorways area Constant 0.2 1.28 1.76 0.96 2.83 11.33 45.85 elasticity 0.3 1.60 2.08 0.96 3.61 13.40 51.28 demand 0.4 1.76 2.24 1.12 4.12 14.94 54.35 0.5 1.92 2.40 1.12 4.64 15.71 56.41 0.6 2.08 2.56 1.12 5.15 16.49 57.96 0.7 2.24 2.56 1.12 5.41 17.00 58.99 Elastic 0.2 1.28 1.92 0.96 3.09 12.36 48.69 exponential 0.3 1.60 2.24 0.96 3.86 14.17 53.07 demand 0.4 2.24 2.40 1.12 4.38 15.30 55.64 0.5 1.92 2.56 1.12 4.89 16.23 57.44 0.6 2.08 2.56 1.12 5.15 16.74 58.73 0.7 2.24 2.72 1.12 5.41 17.26 59.51 Linear 0.2 1.28 1.92 0.96 3.09 12.36 49.20 demand 0.3 1.60 2.24 1.12 3.86 14.43 53.58 0.4 1.92 2.40 1.12 4.64 15.51 56.16 0.5 2.08 2.56 1.12 4.89 16.49 57.70 0.6 2.24 2.56 1.12 5.15 17.00 58.99 0.7 2.24 2.72 1.12 5.41 17.52 60.02 Average 1.9 2.4 1.1 4.4 15.2 55.2 Standard 0.4 0.3 0.1 0.9 1.9 4.2 deviation MCC in cents per pcu-mile 81.9 244.2 784.4 Source: Own calculations, see text

We also obtained potential charges for each town (eighteen for each area) and each demand and elasticity assumption. We found that although different, they were quite similar across demands and elasticities assumed. We chose the constant elasticity demand and assumed a point elasticity of 0.2 because this would give the highest charge. As a starting point, no charging authority would have any basis to introduce a charge higher than that. Once the value of the charge was computed, it was converted into a cordon toll, by multiplying the charge in cents per pcu-mile by the average miles traveled (AMT) within the charged area. The advantage of a cordon toll over distance charging is that it is transparent as drivers know the total price before entering the priced area. It is also simpler to implement and operate. The exercise was done for the three areas in each of the five towns. Charges in cents per pcu-mile and tolls in US dollars to cross the cordon are presented in Table 7.

9

Table 7: Charges in cents per pcu-mile and tolls in US dollars at 1998 prices for different areas in five English towns Area under study Town Includes motorways Excludes motorways Charge Toll Charge Toll Cents per $ to cross Cents per $ to cross pcu mile the cordon Pcu mile the cordon Cambridge 111.8 4.44 213.81 4.52 York 70.07 2.87 110.77 2.06 Norwich 40.44 2 133.95 2.08 Lincoln 115.4 3.73 123.65 3 Exeter 53.84 2.74 123.65 2.61

Central area Charge Toll Cents per $ to cross pcu mile the cordon 311.7 3.68 293.66 2.37 229.26 1 329.73 3.69 298.82 2.23

Source: Own calculations, see text Note: Point elasticity at the original level of traffic: 0.2, demand function assumed: constant elasticity

It could be argued that the only variable considered is number of trips. Drivers have the option of making the trip or not making it during the time period considered, but they do not have the option of changing route to avoid paying the charge. We are at present using SATTAX, a batch file procedure developed at the Institute for Transport Studies at Leeds University, which can be added to SATURN in order to simulate road charging (Milne and Van Vliet, 1993). The program has a facility that allows an elastic demand response, and uses the SATURN elastic assignment algorithm, SATEASY. With this elastic assignment modifications in demand patterns can be assessed. Welfare changes can be estimated for different toll levels. The optimal toll would be that for which the change in social welfare is maximum, as opposed to the one computed using the standard static model which does not consider any change of route on behalf of the drivers. Preliminary results indicate that this efficient toll is lower than the tolls presented in Table 7, roughly in the order of one half (Santos, Rojey and Newbery, 2000). In other words, when drivers are allowed to change route, part of the congestion can be relieved by switching traffic to other less congested areas, and this can be achieved with a toll about half as high as the one necessary to reduce total traffic in the whole of the town. The demand elasticity with respect to road charges is not known. Road charging has not been applied in the UK, except for a few experiments and therefore there are no data available to assess the response of drivers. Given the need to make an assumption on elasticity values, these should be restricted within a close range that is reasonably justified. Taking into consideration studies such as Goodwin (1992), Oum et al (1992), Fowkes et al (1993), Department of the Environment, Transport and the Regions (1998), a sensible range of values would be 0.2 to 0.7. Errors in estimating the efficient charge will not have a major effect on the efficiency of the system. The elasticity of demand is critical in the decision of whether to introduce road pricing or not. A good starting point then would be to introduce road pricing schemes only in places where even with errors in the elasticity values assumed, the gains were large enough to have a benefit-cost ratio greater than one. The way to guarantee this is to assume the lowest possible elasticity, which is what we have done.

10

Would road pricing be an efficient solution in England? Road charging is not the only way to manage traffic demand. There are other ways of restraining demand such as pedestrianization, control in vehicle use, controls on vehicle ownership, physical measures of traffic restraint, public transport improvement, parking charges, etc. None of those measures however would internalise the external costs of congestion. In an ideal world marginal costs and benefits should be equal to achieve an efficient situation. By introducing a toll and making trip makers face the true costs of their trips, only those for which the marginal benefit is greater than or equal to the marginal costs, including the toll they would have to pay, would remain. The rest would be tolled off and would probably switch to other modes, change departure time or destination or decide not to make the trip at all. In the model we use in this study, drivers have the option of making the trip or not making it. The tolls computed are the ones that would be necessary to reduce total traffic demand up to the efficient level, given by Q e on Figure 1. However, on-going research shows that when change of route is modeled, it is not necessary to reduce the total level of traffic on the network up to Qe. A reduction in the DWL can be obtained by simply re-assigning traffic to less congested routes. A lower toll may be enough incentive for drivers to change route. Preliminary estimates show that such tolls would be in the order of one half those presented in Table 7. If deadweight loss is reduced, social welfare is increased. Any step towards an increase in social welfare is a step towards economic efficiency. Only when the costs of implementing road pricing were higher than the reduction in deadweight loss the scheme would not lead to any net gain in social welfare. In those cases, road pricing would not be an efficient solution. A cost-benefit analysis is currently being conducted to determine whether cordon tolls are worth implementing in these towns or not.

Political acceptability The political acceptability of a road pricing scheme is not a minor problem. Nevin and Abbie (1993) summarise the results of a survey of eleven historic British cities and conclude that the major obstacle to introducing road pricing is the political acceptability. The Lex Report on Motoring (Lex Services, 1998) based on two surveys carried out in England in 1997 suggests that while motorists admit congestion and pollution are the two main problems related to car use, they do not perceive restriction on car use or increased taxation as a solution. The survey found that although road pricing would be opposed by majority of drivers, it was still perceived as effective in reducing congestion. The few motorists that would favour such a policy would do so as long as it were implemented in the form of charges rather than taxes, and the revenues stayed in the transport sector. Allocation of revenues can have an important influence on the political acceptability of a road charging scheme. Goodwin (1989, 1990) suggests that revenues could be split in three parts and allocated equally to the following categories: (a) reduction in existing taxes or increase in social expenditure, (b) construction of new roads, improvement of existing ones or improvement in the standards of maintenance of the road infrastructure, (c) improvement in public transport services. He admits that ‘a third’ is somewhat arbitrary but defends it on the grounds that it is a clear division easy to understand and

11

also represents a good starting point to reach an agreement. Small (1992) also proposes a split of revenues in three equal amounts. He defends it on the basis that the program would more than fully compensate the majority and would promote general social goals, all related to the transport sector. The categories of expenditure to which he would allocate the revenues are: (a) monetary reimbursement to trip makers, (b) replacement of general taxes presently used to fund transportation services, (c) new transportation services. Although it may fit in the US taxing framework, option (b) is not suitable for the UK. In the UK road taxes finance other sectors. Road taxes are not earmarked, they go to the Treasury which in turn prepares the budget every year. Road tax revenues in the UK have historically been higher than total road expenditure, but what is worse, the gap has been increasing over the last few years (Newbery and Santos, 1999). For the UK case, option (b) could be ‘the replacement of some part of current road taxes by road user charges’ (Newbery and Santos, 1999). This would also make congestion charges more acceptable to the public that might be prepared to pay them as long as other road taxes were reduced or eliminated. According to the results of the survey carried out by the National Economic Development Office (NEDO, 1991), road charges would be acceptable to the majority (70 per cent of all groups) if the revenues replaced the revenues of other taxes and/or they were invested on roads or public transport. Ison (2000) conducted a survey in which, between other things, he compared public support for urban road pricing before and after the question relating to the use of revenues had been asked. He found that only 11.3 per cent of the respondents viewed road pricing as acceptable (either totally or fairly). Responses changed when the revenues were spent on specific policy options. Road pricing then was perceived as more acceptable (either totally or fairly), with 54.6 per cent of the respondents showing a positive attitude.

Conclusions The model SATURN was used to simulate and assign traffic and estimate the social costs of congestion in five English towns, together with potential road charges. The social costs of congestion were measured by the deadweight loss and computed as the area between the marginal social cost and the inverse demand curves between the actual and efficient levels of traffic. Potential road charges were computed as the difference between average private and marginal social costs at the efficient level of traffic. The exercise is done for three demand functions: linear, elastic exponential and constant elasticity, and five point elasticity values at the original level of traffic: 0.2, 0.3, 0.4, 0.5, 0.6 and 0.7. Three cordons were used in the simulations. The first and second ones followed approximately the contour of the motorways and trunk roads surrounding the town, including and excluding them respectively, and the third one surrounded the central area of each town. Results show that DWL and MCC in cents per pcu-mile increase as the cordon is reduced and, consequently, encircles more congested areas. Annual DWL may decrease or increase as the encircled area decreases. Since annual DWL depends on the product of DWL in cents per pcu-mile and total VMT, it may actually increase if DWL grows more rapidly than VMT decreases. Finally, for a given cordon, DWL and annual DWL increase as the absolute value of the elasticity increases.

12

We also obtained potential charges for each town and found that although different, they were quite similar across demands and elasticities assumed. It could be argued that the only variable considered is number of trips. Drivers have the option of making the trip or not making it during the time period considered, but they do not have the option of changing route to avoid paying the charge. On-going research shows that when change of route is modeled, it is not necessary to reduce the total level of traffic on the network up to Qe. A reduction in the DWL can be obtained by simply reassigning traffic to less congested routes. A lower toll may be enough incentive for drivers to change route. Preliminary estimates show that these tolls would be in the order of one half those estimated in this study. Only when the costs of implementing road pricing were lower than the reduction in deadweight loss the scheme would not be worth implementing. Otherwise, road pricing would be an efficient solution. Finally, the political acceptability of road pricing would have to be assessed. Revenue allocation can have an important influence on the political acceptability of a road charging scheme. Road tax revenues in the UK have historically been higher than total road expenditure. If part of the road pricing revenues were used to replace at least part of current road taxes congestion charges would be more acceptable to the motoring public. Adding road charges to the already excessive road taxes instead of replacing part of the fuel duties and/or vehicle licences with congestion charges will increase the general discontent of drivers and almost ensure failure of the scheme. Only re-structuring the English road charging system can lead to the possibility of charging for congestion.

Acknowledgements Support from the Economic and Social Research Council (ESRC) under Grant R000223117, ‘Road Pricing and Urban Congestion Costs’, from Fundación Antorchas (Argentina) and from the Department of the Environment, Transport and the Regions (DETR), under Contract N° PPAD 9/99/28 is gratefully acknowledged. Any views expressed in this paper are not necessarily those of the ESRC, Fundación Antorchas, DETR or University of Cambridge. The author is indebted to Prof. David Newbery for invaluable guidance and help. All mistakes that survived his corrections are the author’s sole responsibility. The author is also grateful to Prof. Dirck Van Vliet from the Institute for Transport Studies at University of Leeds for his kind patience in clearing her doubts about SATURN. Thanks are also due to Markus Kuhn of the Computer Lab at Cambridge University for software provision to run SATURN from a batch file and for his support and advice when it steadfastly refused to run. The author is also grateful to James Lindsay, from WS Atkins, who spent time solving software problems at short notice. Provision of data by the following local authorities and consultancies working for them is gratefully acknowledged: Director of the Environment and Transport Department of Cambridgeshire County Council, WS Atkins, Lincolnshire County Council, Symonds Group, York City Council, Norfolk County Council and Devon County Council.

13

References Department of the Environment, Transport and the Regions (1998), An Approach to Estimating the Welfare Costs of Congestion through the NRTF 1997 Framework, DETR, London, April. Flowerdew, A. D. (1993), Urban Traffic Congestion in Europe: Road Pricing and Public Transport Demand, Research Report, The Economics Intelligence Unit, London, May. Fowkes, A. S., Sherwood, N. and C. A. Nash (1993), Segmentation of the Travel Market in London: Estimates of Elasticities and Values of Travel Time, Working Paper 345, Institute for Transport Studies, University of Leeds, Leeds, May. Goodwin, P. (1989), The Rule of Three: A Possible Solution to the Political Problem of Competing Objectives for Road Pricing, Traffic Engineering and Control, 30 (10), October, 495-7. Goodwin, P. (1990), How to Make Road Pricing Popular, Economic Affairs, 10, JuneJuly, 6-7. Goodwin, P. (1992), A Review of New Demand Elasticities with Reference to Short and Long Run Effects to Price Changes, Journal of Transport Economics and Policy, 26 (2), 155-169. House of Commons (1999), The Transport Bill, http://www.parliament.the-stationeryoffice.co.uk/pa/cm199900/cmbills/008/1000008.htm Ison, S. (2000), Local Authority and Academic Attitudes to Urban Road Pricing: A UK Perspective, 79th Transportation Research Board Annual Meeting CD-Rom, Washington DC. Lex Services (1998), Lex Report on Motoring: Driving for the Future, Lex Service PLC, Bucks, January. Milne, D. and D. Van Vliet (1993), Implementing Road User Charging in SATURN, ITS Working Paper 410, Institute for Transport Studies, University of Leeds, Leeds, December. National Economic Development Office (1991), A Road User Charge - Londoner’s Views, Report on Survey Findings, Harris/NEDO, London. Cited in Flowerdew (1993). Nevin, M. and L. Abbie (1993), What Price Roads? Practical Issues in the Introduction of Road User Charges in Historic Cities in the UK, Transport Policy, 1 (1), 68-73. Newbery, D. and G. Santos (1999), Road Taxes, Road User Charges and Earmarking, Fiscal Studies, 20 (2) 103-132.

14

Oum T. H., Waters II W. G. and J. S. Yong (1992), Concepts of Price Elasticities of Transport Demand, Journal of Transport Economics and Policy, 26 (2), 139-154. Santos, G., Rojey, L. and D. Newbery (2000), Traffic Management with Congestion Charging in different English Towns, Mimeo, Department of Applied Economics, University of Cambridge, Cambridge, July. Small, K. (1992), Using the Revenues from Congestion Pricing, Transportation, 19, 359381. Van Vliet, D. and M. Hall (1997), SATURN 9.3 User Manual, The Institute for Transport Studies, University of Leeds, Leeds.

15