Designing Efficient Local Road Pricing Schemes
TAG Unit 3.12.1

February 2007

Contents

1. Introduction

2. Objectives and constraints
    2.1 Introduction
    2.2 Reduction in congestion
    2.3 Environmental objectives
    2.4 Social inclusion
    2.5 Local objectives
    2.6 Scheme efficiency

3. Design options
    3.1 Introduction

4. Initial option selection
    4.1 Introduction
    4.2 First stage of design
    4.3 First best pricing
    4.4 Second best

5. Analysing the marginal costs of congestion
    5.1 Introduction
    5.2 The analysis
    5.3 Using select link analysis in cordon design
    5.4 Application to distance based charges
    5.5 Additional analyses

6. Level of charge

7. Practical issues and other considerations
    7.2 Different user groups
    7.3 Road pricing as part of a package
    7.4 Local authority boundaries
    7.5 Economically sensitive areas
    7.6 Highway Agency roads

8. Scheme costs
    8.1 Introduction
    8.2 Period of operation
    8.3 Contingent benefits

9. Detailed scheme design

10. Further information

11. References

12. Document provenance

Annex A. Estimating marginal external congestion costs

1. Introduction

1.1.1 This TAG Unit provides guidance on the design of local road pricing schemes. This covers location, price and periods of operation.

1.1.2 An overview of scheme design issues, including modelling and appraisal issues can be found in Introduction to Modelling and Appraisal for Road Pricing (TAG Unit 2.12). Further detailed guidance for analysts may be found in:

• Modelling for Road Pricing (TAG Unit 3.12.2), which provides advice on the modelling of road pricing schemes;
• Appraisal of Road Pricing Schemes (TAG Unit 3.12.3) which provides guidance on the issues arising when appraising road pricing schemes; and
• Social and Distributional Impacts of Road Pricing Schemes (TAG Unit 3.12.4) which provides guidance on how to assess the social and distributional impacts of road pricing schemes.

1.1.3 This Unit represents the current state of knowledge. This is a rapidly developing area where we are likely to learn from further development in modelling and design approaches and also from the practical implementation of road pricing schemes. This Unit will be revised and updated as further information becomes available or in light of any comments received during consultation and in the application of the Unit.

1.1.4 This unit predominantly considers scheme design in terms of the location of the area to be charged and the structure and level of prices. Scheme design will in practice cover a much broader set of issues around the types of technology, the capital and operating costs, and complementary measures, as road pricing will generally be part of a wider transport and economic package. All these elements are fundamental for a successful scheme. The emphasis in this Unit on the level and location of prices is that there is strong evidence that these matter. Getting the design right is critical for a scheme to deliver good value for money. In practice this means real world benefits in terms of faster and more reliable journeys, improved economic performance and greater public acceptability and business support.

1.1.5 The analytical requirements in these Sections recognise that modelling resources will vary between promoters and that the development of models capable of modelling road pricing will take time. It is also recognised that in some cases, such as the network of a small town, road pricing may not be the most cost effective option for demand management. The core requirement is that an effective analysis of where and when congestion costs occur is undertaken, so that road pricing delivers good value for money. To make the initial analysis cost effective, allowance is made for strategic option appraisal to be undertaken with a less than full suite of fit for purpose models, with appropriate level of detailed scheme design undertaken once the models have been developed. For example, an initial analysis of where and when congestion costs occur can be undertaken using a current highway assignment model. Potential target areas can be selected and the potential effect on at-risk groups can be assessed using various available survey data.

1.1.6 For the full business case analysis the models will need to be fit for purpose and consistent with TAG Unit 3.12.2. It is recommended that the scheme design starts with an initial assessment of locations that capture traffic contributing to congestion. The location and price can then be adjusted to take account of public acceptability issues, distributional and equity issues, scheme costs and financial sustainability.

1.1.7 In much of the following guidance, the design techniques are focused on event-based charges such as a charge for crossing a cordon or for driving within an area as opposed to distance-based charges. There is no presumption against a distance-based charge. However, it is apparent that event-based schemes are currently more technically feasible than a distance-based scheme, in that there are some urban areas with such operational schemes.

1.1.8 The key initial design step is to determine the options for the location of a charge area. Whether the final scheme involves an event-based or a distance-based charge, the area of operation will be important. It will help to determine scheme costs locating equipment and other infrastructure needed for operation, which trips are likely to be charged, any potential boundary impacts, etc. The results of the initial analysis will then focus on the location of a boundary or boundaries, where design issues are generic across different types of schemes. As the results from more research and evaluation becomes available, including for distance based charging, this will be incorporated into guidance.

1.1.9 Sections 2 and 3 set out the objectives and design options. Section 4 sets out the initial approaches to scheme design. Section 5 sets out a way to design efficient scheme locations that help to maximise efficiency. Section 6 sets out how to find the efficient price or set of prices for a given location. Sections 7 to 8 cover practical issues and scheme costs. Section 9 covers detailed scheme design, refining options and choosing preferred options.

2. Objectives and constraints

2.1 Introduction

2.1.1 Scheme objectives and constraints will determine scheme design and need to be clearly set out.

2.1.2 The main objectives will be a reduction in congestion and in most local road pricing schemes this will be through a significant reduction in traffic levels during the peak time in the central or most congested part of the area. A second objective may be the raising of net revenues over and above the operational costs of the scheme Reducing environmental costs could be a significant objective or a constraint on scheme design, in particular location. Distributional and equity issues will act as a constraint on scheme design, though a package including complementary measures may mitigate this.

2.2 Reduction in congestion

2.2.1 A reduction in congestion will provide time savings and improved journey reliability for road users and these will be the main benefits. Reducing the number of trips into a congested area through road pricing will lead to lower journey times for cars in the area. This impact may extend along entire routes, lessening congestion in areas which may not be covered by road pricing. Further, journey time changes and increased demand for public transport may result in increased frequencies with positive indirect impacts on all users of public transport. However, careful scheme design is needed to ensure that diversion does not increase journey times on non-priced routes that could offset time savings on routes that are priced.

2.3 Environmental objectives

2.3.1 The impact of road pricing on the environment can be complex. Reduced traffic flows at peak times will reduce emissions and other environmental costs. However, trips may be made at other times of the day so that total traffic reduction is less. Additionally, it is likely that schemes will induce re-routeing and this will affect environmental costs. For example, if traffic diverts from less populated areas to more populated areas environmental costs will increase and if traffic is diverted to longer routes environmental costs will increase. The overall impact will depend on the reduction in road traffic.

2.4 Social inclusion

2.4.1 Reducing flows of traffic using a demand management tool such as road pricing is likely to affect different social groups in a different manner. Scheme design will need to identify at-risk groups and consider how any negative impacts on these groups could be minimised or mitigated. This could be either through the design of the scheme, the use of any net revenues or through complementary transport improvements.

2.5 Local objectives

2.5.1 Local objectives need to be identified. Road pricing could increase the costs of car access to hospitals, schools, etc. and economically sensitive areas such as regeneration areas for some; complementary measures could reduce the costs of access by public transport.

2.6 Scheme efficiency

2.6.1 An overriding objective is scheme efficiency given objectives and constraints. Schemes with a positive net present value (NPV) will generally be high value for money and the level of NPV will depend on the other scheme objectives and constraints. Clearly there will be trade-offs between objectives. For instance, reductions in the level of road traffic will impact on revenues; inducing longer routes will impact on environmental costs.

2.6.2 It is recommended that objectives should not be set out in terms of specific targets for reduction in congestion or the raising of net revenues at the beginning of the process. The reason for this is that the extent of reduction in congestion and the amount of revenue should be established in the process of careful scheme design.

3. Design options

3.1 Introduction

3.1.1 There are many elements to scheme design. These are:

• Type of scheme - event or distance based charge;
• Location of boundaries;
• Level of price;
• Other elements including pricing by class of vehicle and time of day; exemptions and discounts; and enforcement and penalties.

3.1.2 The combinations of the various elements can give numerous possible schemes. Initial scheme design will necessarily focus on the first two elements of a scheme, identifying the type of scheme and the location of the area. This will help to determine scheme capital and operating costs, locating and quantifying the need for equipment and other infrastructure needed for operation, which trips are likely to be charged, any potential boundary impacts, etc. The pricing schedule in a scheme will be a third important component. Details on types of pricing schemes, structures and charges are set out in Section 2 of Introduction to Modelling and Appraisal for Road Pricing (TAG Unit 2.12).

3.1.3 It may not be possible to model all of the different types of road pricing schemes. The level of complexity will be an issue ranging from a flat price single cordon to a multiple price distance based scheme and the ability to design each type will depend on modelling capability (see Modelling for Road Pricing (TAG Unit 3.12.2)).

4. Initial option selection

4.1 Introduction

4.1.1 The advice in this section reflects research undertaken about efficient scheme designs and the methods used have been broadly categorised into first stage analysis, benefit maximising ('first best') and second best. First-stage and second best are tools to help in the scheme design. As part of this process, the benefit maximising design gives some initial benchmark against which schemes can be compared and is a useful tool for analysis, as well as being a good starting point for scheme design.

4.2 First stages of scheme design

4.2.1 Scheme design will be an iterative process, combining data analysis and modelling work, with expertise from those involved in implementing the road pricing scheme. It is likely that some early analysis, using readily available quantitative and qualitative information, will help to identify possible areas to be covered by road pricing. Such initial design will begin with a judgement about acceptable locations, such as a recognisable boundary like a ring road, identifying potential options for the area to be covered by a scheme. A selection of options should be identified as designing schemes on the basis of judgment and natural boundaries is only likely to provide positive change in social surplus by chance. This section suggests some straightforward analyses that can be used to sift proposed charge areas.

4.2.2 Initial analysis will follow the same structure as general scheme appraisal, starting by identifying the congestion problem faced by the area to be charged. This will involve estimating the peak flow of traffic into the area and the associated congestion. Most scheme designers would have 'congestion maps' available to them, highlighting which roads are close to capacity. They may also have marginal external cost maps (described later). In some circumstances, it will be possible to model the benefits maximising prices (see later) which will estimate the level of traffic flows along links when an efficient road pricing scheme is in operation. It is useful for scheme designers to use evidence to identify what level of flow may be associated with successful lowering of congestion. As noted earlier, it is unlikely that a precise estimate for a particular area is possible without modelling.

4.2.3 The judgemental approach in scheme design could focus on analysing charge areas. For a suitable range of options the pattern of trip origin/destinations, focussing primarily on the most congested times of day, should be analysed. As this analysis will primarily identify an area for charging, it would be easiest to assume a cordon or area wide user charge will circle an area of congestion. By varying the boundaries, the impact both of the location and of the size of the charged area can be examined. A selection of potential areas may be identified varying the size covered by the charge. Where an option involves multiple areas, analysis should be undertaken on individual areas and on the areas combined.

4.2.4 It is often suggested that there is a good case for allowing alternative routes, such as locating cordons within ring roads. Such an alternative route will allow journeys which do not start or end in the congested areas to avoid being charged. However, this would only be of benefit if the ring road is less congested than the area within. Alternatively administrative or natural boundaries may provide a boundary for a possible option. At this stage, it would be best to provide a selection of diverse options to ensure that the data analysis can be generalised to a range of schemes.

4.2.5 Origin-destination datasets covering an area sufficiently large to include a high proportion of trips into the potential charge area may be available to scheme designers. Such data, combined with information on the modal split of journeys, would provide some high-level indications of the flows of traffic that would be affected by a potential charge area. Given the charge area to be analysed, origin-destination information can be used to identify the number of trips according to a range of disaggregations, such as journey purpose and time of day.

4.2.6 A second set of analyses should be undertaken taking into account the potential charge area. A useful disaggregation is whether none, one or both trip ends are in the proposed area if an event based scheme is proposed. A further analysis would focus on trips that do enter the charge area, identifying how far into the charge area their routes go, because trips that end close to the boundary would be more likely to avoid the charge area by stopping short of the charge area. Analysis of the other end of these trips will indicate which radial routes or corridors into the congested area are used by these trips.

4.2.7 The second category of analysis should provide a particularly useful set of breakdowns quantifying the extent to which traffic would pass through a charge area screen-line. In the busiest times, such as the AM peak, trips with only the destination end in the potential area would be affected by the charge. The behaviour of these travellers would then be impacted by the road pricing scheme. For a trip with no trip end in the charge area, the effect of road pricing can be to deter the journey crossing through the area, with it re-routed around the charge area, if such a route is available.

4.2.8 For trips where both ends are in the proposed charge area, the incentives will be different if the scheme is a cordon or an area charge. For a cordon, such a trip would be unaffected by road pricing as no charge would be incurred. In an area wide congestion charge, the scheme designer might consider whether such a trip would be allowed some resident discount.

4.2.9 Figure 1 summarises this breakdown. The scheme designer will provide an early analysis of the level of traffic that may or may not be affected by the charge area. It would then be possible to combine this information with some assumptions about likely behaviour to give an indication of whether the area option under consideration could reduce traffic sufficiently to reduce congestion.

Figure 1: Breakdown of scheme impacts by orgin and destination

* Note. If the cordon or area is small there may be an option to park and walk.

4.2.10 The origin-destination information may be improved by other commonly available datasets. In the busiest times, it would be possible to analyse information on trip destinations. For example, in the AM peak, the location of employment is likely to be the determinant of a significant proportion of destinations predicting the number of journeys into congested areas. Other information may be available for likely destinations, such as places of education possibly from work on accessibility planning, to provide the density of trip destinations in the scheme area.

4.2.11 The important result from such a high level analysis is that the number of trips likely to be deterred by a charge will be identified for a range of charge areas. In an area-based charge, the boundary of the area is the determinant of whether a trip will be charged. Some general results can be given. A charge area that is small is likely to have trips ending short of the charge area. This would mean the decision to travel would be unaffected by the charge for a large number of trips and the effect on congestion of the road pricing may be low. Similarly, an area that is very large may not affect enough trips that are destined for a central congested area. The effect may be to reduce trips, but that these trips did not contribute significantly to the congestion problem in the area.

4.2.12 It should be noted that a careful initial analysis replicates some of the steps in the 'select link analysis' that is described below. That approach automates the processes that are described here and so offers a faster and more robust means to undertake this initial option sifting and provide designs that better meet the efficiency objective. The main advantage of the first stage approach is that it allows an intuitive analysis of the most commonly used transport dataset, namely the origin-destination patterns. While the analysis will not be as robust, it will highlight the geography of congestion in a local area, possibly involving a wider range of the scheme design team. It may also indicate any data shortcomings or other analytical issues which may be corrected at this early stage in scheme design.

4.2.13 However, it will not be possible to design an efficient scheme if no account is taken of the relationships between congestion costs, charges and willingness to pay. If the aim of road pricing is to change behaviour this needs to be closely linked to the costs of congestion in terms of the willingness to pay for a reduction in congestion. The following sections set out how the initial first stage analysis can be significantly improved on with the aim of meeting the efficiency objective set out above.

4.3 First best pricing

4.3.1 First best pricing sets charges at the level where marginal social costs equal the willingness to pay for road trips. It is an analytical tool providing three useful results from a transport modelling system. The first is an estimation of a benchmark against which second best schemes can be judged. This is because the gross benefits (the sum of changes in time savings, user charges, scheme revenues, tax revenues and any other monetised costs except scheme costs) will be maximised within the model. Secondly, the modelled prices will indicate the marginal external costs of a road user at this maximum. The third is that it provides the basis for a more detailed examination of congestion costs in particular by testing what happens in a particular transport model when road pricing is introduced.

4.3.2 Figure 2 details some of the terminology used and provides the theoretical background to first best pricing. The figure indicates the generalised cost of a trip at various levels of traffic volume. The right of the graph, with rising costs, is where congestion means an additional trip is increasing costs both for the trip-maker (private costs) and imposing costs on other users of the road (external costs).

Figure 2: First best benchmark diagram

4.3.3 Without any charge trips will be made using a particular link up to the point where average private costs equal the willingness to pay or demand for a trip, N_A in the diagram above. At this point there will be excess demand for road space as marginal social costs exceed demand. This arises as road users do not take account of the costs they impose on others when making the decision to make a trip or not. These costs include the increased journey time that others will face (marginal external congestion costs) and environmental costs. There may also be other resource costs such as increases in fuel consumption, road wear and tear, and parking resource costs that may not be considered.

4.3.4 The efficient level of traffic would be at N_o where marginal social costs equal willingness to pay. Setting a price equal to the difference between marginal social costs and average private costs would reduce demand for trips to a socially efficient level. This is shown as the dark blue line in the Figure 2 above.

4.3.5 Road pricing provides a means to internalise these external costs to enable road users to make more efficient decisions. To maximise benefits, prices can be applied that exactly match these external costs at the point where demand equals marginal external costs plus generalised costs. It is important to note the external costs with road pricing in place will be lower than the external costs before road pricing is introduced.

4.3.6 Depending on the model used in the analysis, this approach could be extended in a number of directions. Prices could be allowed to vary by each link, by class of vehicle, and by any other variable that had an impact on the travel time of all other users (and that all congested roads have road pricing).

4.3.7 Clearly this would be complex to impose and, though it maximises the benefits of the road pricing scheme for society, it could conflict with other objectives. For example, the scheme costs could be much higher than alternatives, which do no not have as high user benefits. Nevertheless, in modelling this scenario, it does provide a useful benchmark against which to gauge schemes options that by necessity will result in lower gross benefits (the sum of changes in time savings, user charges, scheme revenues, tax revenues and any other monetised costs except scheme costs). It will be explained below why this is an important part of the analysis in scheme design.

4.3.8 A 'first best' analysis provides the backdrop against which to design scheme options that take account the objectives and constraints.

4.4 Second best

4.4.1 Second best takes account of scheme objectives and constraints and attempts to maximise benefits given these. To some extent, scheme designers may be able to start from the optimal, benefit maximising scenario. However, the location of the scheme, times of operation and prices will differ from this recognising that other objectives in scheme design will also have to be taken into account.

4.4.2 The gross benefits of a second best scheme will diverge from the benefit maximising scenario because of a range of reasons. For example, the charging area will cover only part of the network. There will be some roads that are priced and some that are not. In the case of first best all links would be priced and any re-routeing would maximise the benefits of using road space (this is a system optimum). If benefit maximising first best prices were applied only to a subset of links, excessive re-routeing to non-priced roads could reduce the benefits. The second best price factors down the first best price to optimise the net benefits of differently priced routes. In other words it is a first best price less an amount to minimise the impact of distortions elsewhere.

4.4.3 In summary, this section has considered the role of first stage analysis and benefit maximising ('first best') analysis in developing a practical 'second best' road pricing scheme design. The first stage sift of information may give a good steer on where to locate the charging area. The optimal, 'first best' analysis will help to refine the location of the charging area. The 'first best' prices estimated will enable the analyst to select appropriate practical 'second best' prices. The practical constraints incorporated in the 'second best' analysis will result in 'second best' prices that are lower than 'first best'.

5. Analysing the marginal costs of congestion

5.1 Introduction

5.1.1 This Section sets out in more detail how the costs of congestion can be identified and used in designing efficient locations. Although it has been noted in Section 1 that this guidance is focussed on event based charging schemes such as cordon or area schemes the methodologies set out could be used to identify prices for efficient distance based charging.

5.1.2 Section 4 set out the importance of identifying marginal costs and pricing accordingly. It also noted that, for practical purposes and for local road pricing schemes, it would be sufficient to identify only marginal external congestion costs (MECC), which are based only on the time delays imposed on all other users. Marginal social costs (MSC) include a wider range of the external costs, such as environmental externalities but also exclude fuel duties. Annex A to this unit provides more detail about the definition of MECC with a method for calculating them in a relatively straightforward manner. In Annex B of Modelling for Road Pricing (TAG Unit 3.12.2) there is a detailed description of the marginal social cost pricing method, where road pricing reflects the full marginal external costs of an additional vehicle, including environmental and other non-congestion external costs.

5.1.3 The calculated marginal external costs - hereafter a general term for both the marginal external congestion costs (MECCs) or using prices based on marginal social costs - will depend significantly on whether or not road pricing has been modelled on the network. Where there is no road pricing, the levels of traffic and associated external costs would, at the margin, be higher. As road pricing is applied, the traffic flows should decrease lowering the marginal external costs, reflecting the reduced congestion. It is recognised that it is difficult for models to estimate the marginal external costs with road pricing and a first analysis may use marginal external costs without any road pricing assumed. Scheme designers are encouraged to discuss with the Department results from such analysis and possible further steps towards an optimal marginal external cost analysis.

5.2 The analysis

5.2.1 Estimation of marginal external costs is important for several reasons. Firstly, they serve as status indicators that describe the current state and trends of congestion. Secondly, changes in marginal external costs provide the basis for cost-benefit analysis that assesses whether individual projects and programs are worthwhile investments. Finally, and possibly most importantly, obtaining accurate marginal external costs is crucial for designing efficient pricing schemes.

5.2.2 Marginal external costs will depend on route choices, feasible alternatives, flexibility in arrival times, perceived costs of all of the choices, journey purpose and willingness to pay or demand for the trip. The perceived costs (or disutility of travel) will depend on values of time and other personal attributes of users of the route. This means that good quality data is needed, not just on the network and levels of service but on the personal attributes of road users.

5.2.3 To get good estimates of marginal external costs a high level of demand segmentation is desirable but there is a trade-off between segmentation and run times and available resources. Even if there is limited segmentation it is essential that a range of willingness to pay by income is represented. Further information on minimum requirements is given in Modelling Road Pricing (TAG Unit 3.12.2).

5.2.4 If modelling capacity is limited at the outset of a study, the estimation of marginal external costs can be undertaken using only a highway assignment model. A method for calculating marginal external costs which focuses on the congestions costs over the network is set out in Annex A of this TAG Unit.

5.2.5 A useful way of presenting marginal external costs is via maps that show the marginal external costs for each link. An example of this is shown below which uses the SATURN package and estimates the congestion component of marginal external costs only.

5.2.6 Figure 3 below shows an example of an analysis of a simplified buffered network using speed/flow analysis. This modelling was undertaken with SATURN in buffered mode to derive the theoretically optimal level of road user charges and the effect on traveller behaviour (May (2002)). Network assignment usually runs to an equilibrium where no trips could reroute to reduce average costs. In the May (2002) analysis, an equilibrium was found where the marginal external costs could not be reduced by rerouting, under conditions of elastic demand. The diagram shows that the main congestion is on the radial approaches to the centre except for one orbital route to the east and one river crossing to the north.

Figure 3: 15 top MECC links

5.2.7 Similar maps can be constructed for MECC without road pricing to identify where the links with the highest levels of congestion are located and these are likely to be those links which are common to a number of routes causing bottlenecks. On these routes road users may not be able to avoid certain links if they are to make the trip at minimum cost. These common links are likely to be the most congested and identification of these common links and the level of congestion throughout the day would a useful starting point in any analysis. At a very basic level maps could be drawn by comparing demand with capacity on major links and selecting those with the highest ratios.

5.3 Using select link analysis in cordon design

5.3.1 The marginal external costs analysis will show which links are highly congested. For cordons and area charges the analysis described in this section will provide an efficient location and thus boundary for the charging area. The method has been developed by the Institute of Transport Studies, University of Leeds that combines this with flows to aid the design of efficient cordon locations using select link analysis. The core of the approach is to carry out an analysis of a limited number of links with the highest MECCs, hence select link analysis.

5.3.2 Select link analysis is an aid to designing an efficient closed pricing cordon by looking at the proportion of flows on "high cost" links which are potentially caught by a charge area. The technique currently assumes that the charge area will be managed as a cordon scheme. However, as the technique is used in locating a charge area, it should offer significant insights for an area-based scheme.

5.3.3 The approach addresses the problem that, whilst congestion tends to get heavier as one approaches urban centres, network topology also exerts an influence on the location of congestion. Consequently, trying to draw cordons that connect or contain the top number of the most congested links (regardless of whether they are the top 10, 20 or 100), is often not possible - the congestion may be heaviest at 2 or 3 non-contiguous places across the network. The select link analysis approach allows a cordon to be designed that captures a large proportion of the traffic that uses the most congested links with a single scheme without necessarily having to price trips directly on the most congested links. Think of it as 'remote pricing'. The rationale is that it is not necessary to include the top links in a cordon but it is important to cover as much as possible of the path flows from those links by the cordon. The hypothesis is that the higher the proportion included is, the greater the gross benefit from the cordon will be.

5.3.4 This method can be used with all assignment models. For the location of boundaries of cordons and area schemes the important issue is the relative levels of marginal costs on each link. Excepting for scale, these relative differences should be the same however marginal costs are estimated. Thus the most congested links can be identified using MECC in the absence of pricing as in Annex A to this TAG Unit, or MECC based prices if the assignment model has a system optimisation facility (see above) or using the more comprehensive marginal social cost approach set out in the annex to Modelling Road Pricing (TAG Unit 3.12.2). Where various options are to be tested using distance based charges it is recommended that the MECC approach detailed in Annex A is used (see Section 5.4 below).

5.3.5 The select link analysis shows where flows using the selected links come from and go to through the rest of the network (In SATURN this can be done using the P1X program). These flows may be used to help locate cordons. A calculation of the proportion of flow on the selected links that is "caught" by the cordon provides a measure of the effectiveness of the cordon.

5.3.6 The steps in the process can be summarised as follows:

1. Estimate MECC or MECC based prices using a buffered network.
2. Select the top X links by price or congestion cost level.
3. Analyse the flows on selected links on routes that include the top X links.
4. Draw a cordon (or set of cordons) that "catches" a high proportion of the flows from the selected links.
5. Optimise the uniform charge level for the selected cordon(s) (see Section 6 below).
6. Calculate the proportion of flows on the selected links that are captured by the cordon.
7. Adjust the cordon(s) identified in step 4 to maximise the proportion of flows on the selected links that are captured by the cordon.
8. Re-optimise the uniform charge level for the selected cordons (see Section 6 below) and compare gross benefits.

5.3.7 This approach combines analysis with the judgement required to take account of the objectives and constraints.

5.4 Application to distance based charges

5.4.1 The above analysis is generally applicable to whatever type of scheme is proposed. For example, for distance based charging it will provide information about marginal external costs that would be vital in considering the appropriate average distance based charges. The level of averaging will depend on public acceptability, the type of distance based scheme in consideration and modelling capability.

5.4.2 When local distance based charges are being considered it is recommended that the MECC approach detailed in Annex A is used. This is because the network effects arising from route and mode choice need to be taken into account to ensure that the appropriate costs are estimated (estimating link costs in isolation may over or underestimate true costs where only a sub-set of links are priced). Also, the use of the MECC approach means that the main external costs will be estimated without the need to take account of fuel duty and other differences between resource costs and perceived costs. Whilst these are important in estimating the optimal link based charge the MECC approach should result in similar prices that would provide a good basis for choosing average prices. The efficiency of the average prices can be examined by calculating the change in gross benefits.

5.5 Additional analyses

5.5.1 It should be possible to use the same approach to address environmental problems. This would entail capturing flows that added most to environmental costs.

5.5.2 For cordons where there is a significant amount of through traffic and there are alternative unpriced routes that are close substitutes, demand within the cordon will fall as a result of less terminating traffic and through traffic re-routing. If some of these alternative routes have similar or higher levels of congestion than the levels of congestion within the cordon before pricing, then there could be a significant fall in overall time savings.

5.5.3 The potential for rerouting could be analysed by using the methods above or by setting a price on links with high MECC his could be assumed to be two thirds of the ex-ante MECC and would give a good indication of the likely levels of rerouting.

5.5.4 It could be that for some routes where there is a high proportion of users with high values of time there is less rerouting and for other routes where there is a high proportion of users with low values of time there is more rerouting.

5.5.5 The analysis above gives a lot of information about the demand for and supply of the network. It is critically dependent on the behaviour of road users and, as such, the level of segmentation that accurately portrays road user characteristics. It will show the level of rerouting that is possible and the consequences of pricing on some routes while leaving alternative routes non-priced.

6. Level of charge

6.1.1 Having designed efficient locations, i.e. scheme options that capture a high amount of congestion, it will be necessary to estimate the efficient charge. For all schemes the level of charge will be critical in delivering benefits and revenues. For cordons the charge that delivers the highest gross benefits ((the sum of changes in time savings, user charges, scheme revenues, tax revenues and any other monetised benefits except scheme costs) can be identified by looking at different charging levels and the resulting gross benefits For example, nine different levels of cordon charge could be used. Illustrative results for such an exercise are shown below.

6.1.2 It can be seen the gross benefits rise as the charges are increased, reach a peak at £1.25 in this hypothetical example then tail off as the charge is increased further. At the optimal level of charge, time savings and the difference between user charges and revenues are maximised. Either side of the optimal charge, the time savings will fall faster than the difference between user charges and revenues. At the extreme if charges are high enough to bring traffic levels down to free flow speeds any increase in charges would make gross benefits negative as there would be no offsetting time savings.

6.1.3 This analysis will show how much benefit is foregone by adopting charges that are lower than or higher than the optimum. There may be trade-off between economic efficiency and public acceptability and any incentive to maximise revenues.

6.1.4 For distance based charges a similar approach can be adopted except that the comparison will be between options with different levels of charges. Similar analysis might consider the trade-offs between overall benefits and the variability of charges for different times and locations.

7. Practical issues and other considerations

7.1.1 There are a number of practical issues that will affect scheme design. These include: impact on different users; ease of describing the charge; scheme design where road pricing is part of a package, local authority boundary issues; economically sensitive areas; and Highway Agency Roads.

7.2 Different user groups

7.2.1 The impact on different groups of users will affect scheme design and public acceptability. Details on the importance of income and the need for segmentation by this and journey purpose are given in Modelling for Road Pricing (TAG Unit 3.12.2). The geographical location of any pricing scheme and the type of traffic could have different effects on different wards or location within and without boundaries, and there will be distributional and equity analyses that need to be made. Details of how to analyse these impacts are given in Appraisal of Road Pricing Options (TAG Unit 3.12.3) and in Social and Distributional Impacts of Road Pricing Schemes (TAG Unit 3.12.4).

7.3 Road pricing as part of a package

7.3.1 Road pricing will generally be part of a wider package of transport and economic measures, and many of these complementary measures may be designed to provide better transport opportunities for groups affected by road pricing. The use of net revenues may also be an important factor in improving the public acceptability and economic impact of pricing schemes.

7.3.2 All these factors will need to be taken into account in the scheme design, and will be highly dependent upon the local circumstances of the scheme. In order to do this the starting point is having the required information on the optimal pricing structures and the distributional implications of the scheme. There will then need to be process of require testing the performance of the overall package with a range of complementary measures, combined with judgements on public acceptability and scenarios on how revenues may be used to maximise the impact and support for the scheme.

7.4 Local authority boundaries

7.4.1 Scheme design will cover the core network and the hinterland. In many cases the area to be charged could be covered by more than one local authority or different tiers of local government.

7.5 Economically sensitive areas

7.5.1 The objectives of efficient scheme design and the development of economically sensitive areas could well conflict. Regeneration areas are closely tied to accessibility that could be affected by a road pricing scheme.

7.6 Highways Agency roads

7.6.1 It is likely that in some schemes the pricing area will cover or impact upon the strategic road network. The close involvement of the Highways Agency in the development of any proposals will be important to avoid conflicts of objectives.

8. Scheme costs

8.1 Introduction

8.1.2 Scheme design should reflect congestion costs and distributional concerns but financial viability will be an issue and this is clearly linked to value for money. The choice of scheme technology and the costs (capital and operating) will have an impact on scheme design in two ways. The technology will determine how variable the prices can be and, thus, the efficiency of the scheme in tackling the congestion problem, while the costs will affect the size of the scheme and the level of net benefits. The most efficient scheme may not be the one with the highest benefits or the best value for money, as cheaper schemes may offer greater net benefits. This would indicate that there should be some flexibility in scheme design and it will be important to estimate the costs of alternative designs. Separate advice on the costs of alternative scheme technologies is also being prepared by the Department and scheme developers should talk to the Department about this issue.

8.2 Period of operation

8.2.1 Financial viability or cost recovery will depend on the period of operation and traffic volumes. Schemes could operate all day or for part of the day. In some situations the operating costs might outweigh the revenues. If the period of operation is short, for example the am peak only, the operating costs could be significant and, with small schemes, outweigh the revenues. Consequently, the technology costs are a determinant of the size of the scheme. Scheme developers should talk to the Department about this issue.

8.3 Contingent benefits

8.3.1 It is recognised that there could be contingent benefits from the adoption of technologies that accelerate the take-up of road pricing, though this will be an issue for the Department to judge.

9. Detailed scheme design

9.1.1 The initial analysis described above will be useful for clarifying the problem, where and when congestion is highest, the potential for rerouting and the transfer of congestion, and the location of at-risk groups. Options for location, charges and time of operation for the type of scheme chosen can be identified but the modelling tools used for the initial option selection may not allow for a sufficiently detailed analysis of demand responses or for sufficient heterogeneity in response. Analysis of demand responses is critical for the selection of options and to help in deciding where complementary measures would best perform either to mitigate adverse affects on some groups or enhance the benefits overall.

9.1.2 Therefore a further level of detailed scheme design will need to be undertaken. In order to move towards a firmer consideration of the options a minimum requirement will be a properly constructed, fully developed and fit for purpose four of five stage transport model that is sufficiently segmented.

9.1.3 The use of fit for purpose models is the only way that scheme options can be appraised robustly. This will identify which of the initial options are worth pursuing and which options could be improved or adjusted to increase gross benefits or to minimise conflicting objectives that may not be apparent using less well defined models. For example, the use of a highway assignment model with elastic demand will not be able to identify the various different demand responses that will depend on the availability of choices and personal preferences, except in a very aggregate sense, and these will impact on scheme design.

9.1.4 For cordons, methods are being developed for selecting and combining links that will provide efficient locations and charges that take account of objectives and constraints. This is done by maximising net benefits depending on the objective function using a combination of links. These combinations are based on the concept of genetic algorithms that have been tested on coarse networks with single user classes. These should enable efficient cordon designs to be made that incorporate objectives and constraints and allow for more efficient judgments to be made. This methodology has yet to be applied to networks with finer zones and multiple user classes.

9.1.5 In the meantime best practice in scheme design should incorporate the analysis identified above, using fit for purpose models, but recognising that an element of judgement will be necessary in designing second best schemes.

9.1.6 As discussed above, a first best solution provides a useful benchmark against which to judge the benefits of second best schemes. Modelling for Road Pricing (TAG Unit 3.12.2) outlines how existing modelling tools may be modified to enable the identification of first best prices. For some modelling packages, the model user may be able to make these modifications, but for others it may be necessary to seek the assistance of the package owner.

9.1.7 A second best solution may be obtained by using the initial analysis of flows and congestion costs to identify areas that capture a number of highly congested links and most of the terminating flows of the congested links that are outside any priced area. Modelling and appraisal will show how travellers will respond to the options, enabling the selection of preferred options that fit best with the scheme objectives and refinements to be made to these options to reduce undesirable responses.

9.1.8 The analysis of marginal external costs and flows should enable a range of locations and prices to be identified for testing. Whether a distance based, cordon, or link based scheme is being considered, the principle of setting a price based on the MECC at the outset should be preserved. Where system optimisation is not available choosing a price that is around two thirds of the estimated ex ante MECC based on income segmented demand can be used to direct analyses of welfare benefits in the search of the best location and price. The subsequent averaging of link prices will depend on public acceptability and modelling capability in terms of testing for value for money. For cordons and area wide pricing this would not be a problem as, unless there are multiple cordons, it is likely that a flat price or stepped price would be appropriate. For distance based prices, testable variations may depend on how much variation in marginal congestion cost there is.

9.1.9 The preferred options should be developed from the initial options that can be developed using assignment models for a first sift, where more complex models are not available. If this analysis shows that road pricing is an efficient means of traffic management, and this will depend heavily on the size of the scheme and the levels of congestion, this would provide the incentive to develop fit for purpose models. Where close to fit for purpose models are available, noting the requirements in Modelling for Road Pricing (TAG Unit 3.12.2), the initial analysis can be undertaken at a more detailed level though for initial sifting some of the stages such as trip generation and destination (and time of day choice) can be switched off to improve run times allowing more options to be tested.

9.1.10 One of the main benefits of going through the above procedures is that efficient schemes will be designed that have the appraisal process explicitly incorporated in the design stages. If followed, the result of the value for money appraisal that follows the guidance in Appraisal of Road Pricing Schemes (TAG Unit 3.12.3) should not be a surprise.

10. Further information

Technical queries and comments on this TAG Unit should be referred to:

Integrated Transport Economic Appraisal (ITEA) Division
Department for Transport
Zone 3/08 Great Minster House
76 Marsham Street
London, SW1P 4DR
E-mail: itea@dft.gsi.gov.uk
Tel: 020 7944 6176
Fax: 020 7944 2198

11. References

A.D. May, R. Liu, S.P. Shepherd, and A. Sumalee (2002) "The impact of cordon design on the performance of road pricing schemes". Transport Policy 9 209-220.

12. Document provenance

This Transport Analysis Guidance (TAG) Unit revises guidance published for consultation in July, 2006.

Annex A. Estimating marginal external congestion costs

A.1.1 In order to account for the interaction between traffic congestion on different links on the network, one has to perform a new traffic assignment and compute MECC on links resulting from changes in speeds on the entire network. This is done by decreasing or increasing the demand for travel between all origin-destination pairs by the same small percentage, one that will not be lost in the assignment process e.g. 5 per cent and run the highway assignment model with the new initial demand. Then, the MECC on link k can be computed as follows:

where t_ks is the travel time for demand segment s on link k, qks is the flow in demand segment s on link k, VOTs is the value of time for demand segment s, and subscripts 0 and 1 denote initial and resulting traffic assignments. To convert into marginal external costs per kilometre these costs will need to be divided by link length.

A.1.2 An important advantage of this method is that it accounts for the interaction of the speeds and traffic flows on the network. It also ensures that the estimated MECC reflect variations by location according to income levels and the mix of journey purposes. In fact, it accounts for all the network effects and is very cost-effective since obtaining a full set of MECC using this method requires only two model runs.

A.1.3 It is important to note that this method will provide the MECC in the absence of pricing. If pricing is introduced, the traffic flows should decease, leading to a reduction in MECC. Therefore, the main use for this approach is the identification of links with high congestion costs to be used in further analysis.

A.1.4 If prices equal to MECC are required, an iterative procedure will be required to estimate the traffic levels and prices that represent equilibrium. Annex B in Modelling Road Pricing (TAG Unit 3.12.2) presents two techniques that have been applied in the Department's National Transport Model to address this problem.