MSA: Road Decongestion Benefits
TAG Unit 3.9.5

August 2007

1. Introduction

2. Deriving Decongestion Benefits

3. Estimates of the Marginal External Costs of Car Traffic

4. Application of Marginal External Costs as Decongestion benefits

5. Further Documents

6. References

7. Document Provenance

8. Annex 11
   8.1 Annex A: Marginal External Costs and Area Type Classification
   8.2 Annex B: Congestion Band Definition

1. Introduction

1.1 Road decongestion benefits should be considered by promoters of transport schemes where significant traffic reductions are likely to occur in moderate to congested situations. In uncongested areas the effects of reduced traffic flows would be minimal, analogous to moving along the flat part of a traditional speed / flow curve. This TAG unit discusses the assessment of road decongestion benefits arising from transport schemes focussing on evaluating the decongestion impacts on schemes which reduce car traffic.

1.2 In some areas the overall journey time effects of traffic flow changes may be small and limited to the peak periods or later years of the scheme. In other cases, the secondary effects on highway journey times of modal shift due to induced traffic must be included in the appraisal. This guidance documents valuation methods for the benefits to other road users of any reduction in car, bus or other non-freight vehicle traffic. Other guidance offers more detail from the perspective of assessing the particular intervention, such as guidance TAG Unit 3.13.2: Guidance on Rail Appraisal: External Costs of Car Use or TAG Unit 3.14.1: Guidance on the Appraisal of Walking and Cycling Schemes.

1.3 The primary difficulty in assessing road traffic reduction benefits is that traffic reductions can often be spread over a large number of highway links. Therefore changes on each link may be small compared with the total flow of vehicles.

1.4 For larger schemes, the approach to the assessment of road decongestion benefits must not be constrained by the need to develop new, or improve existing, modelling tools.

1.5 However, in exceptional circumstances and for smaller schemes, more simplified approaches to estimating decongestion benefits may be appropriate. Promoters should discuss and agree with the Department the use of simplified approaches. In such cases the reasons for adopting less complex approaches should be outlined.

2. Deriving Decongestion Benefits

2.1 Decongestion benefits value the journey time savings and other costs and externalities due to the removal of a vehicle kilometre from a road. Where traffic is added to a road such a calculation will value the congestion costs of the additional traffic.

2.2 Where scheme costs and impacts are large (for example, light rail or guided bus proposals), decongestion benefits should be assessed using the fully specified multi-modal modelling approach. This approach includes modal choice and thus ensures that levels of mode switching and highway journey times are mutually consistent. Significant highway decongestion may also lead to induced traffic as a result of trip generation and changes in destination choice. The need to model these responses (either directly or using elasticity methods) must be considered on a case by case basis.

2.3 Where scheme costs and impacts are not large, the range of approaches set out below should be considered. The methods require a robust estimate of the number of car, bus or other road vehicle trips that will be removed or added as a result of the change due to the scheme. The final two methods require a highway assignment model. The methodology that promoters adopt must be agreed with the Department:

• The Department's National Transport Model (NTM) has been used to estimate the marginal external costs of adding car traffic to the road. These costs have been disaggregated by the type of roads, level of congestion and area type. An approach can be employed combining these congestion costs with estimates of changes in car traffic. More detail about these estimates of congestion costs is provided in section 3 of this unit;
• Manual reduction of flows on the affected highway links. As this is a simple link-based approach, the output from COBA can be analysed to determine the average cost per vehicle at different flow levels. This approach should only be used where the number of highway trips removed is small and the routing of highway trips can be assumed to be unaffected;
• Manual reduction of trips for the affected cells of the highway trip matrix. Following this, a highway assignment model should be applied and benefits can be assessed using TUBA. This method should be used where re-routing of highway trips is expected, but secondary induced traffic effects can be ignored;
• Where changes in highway journey times are significant and these benefits become a significant proportion (say, about 10%) of the transport economic efficiency benefits, induced traffic should be taken into account via an augmented application of the method discussed in the preceding bullet. Trips in affected cells of the highway trip matrix may be manually reduced. However, when applying the highway assignment model, elasticities should be included to cater for induced traffic. Further guidance on the use of elasticities to estimate induced traffic is given in Appendix 1 of Variable Demand Modelling - Key Processes (TAG Unit 3.10.3). TUBA should be used to assess the decongestion benefits.

2.4 Where the proposed scheme has a significant effect on the capacity of the highway network (including proposals for targeted traffic management measures intended to minimise induced traffic or, for example, where on street running of light rail is proposed), this should be reflected in changes to the highway model.

2.5 Where highway assignment models are used, they must be based on a detailed network, with modelling of flow dependent link speeds and junction delays. These models should be used iteratively to attain convergence of results such that flow and travel times on links do not change greatly between iterations. Advice on the requirements of road traffic assignment models is given in Traffic Appraisal in Urban Areas, The Design Manual for Roads and Bridges, Volume 12.2.1 and Modelling (TAG Unit 3.1).

3. Estimates of the Marginal External Costs of Car Traffic

3.1 The Department for Transport's National Transport Model (NTM) has been used to calculate the marginal external costs of car traffic. The marginal external cost (MEC) is the cost imposed on society by adding a marginal vehicle to the road. The calculation of the MEC aggregates across all other users of the road the change in the total delay and the total change in vehicle resource costs. In addition, there are other changes in external costs such as environmental externalities. The annex to TAG Unit 3.12.2, Modelling for Road Pricing, gives a detailed description of the various concepts.

3.2 This section describes how marginal external costs have been calculated using the NTM. It is a multi-modal model which includes 6 modes of transport - car driver, car passenger, rail, bus, walk and cycle. The model is composed of a series of sub-models, three of which are applied in iteration to produce the main model outputs. More information on the NTM is available on the DfT's website. The particular model runs used to calculate the marginal external costs were used in DfT (2005) and more information on the underlying assumptions are on the DfT's website. The assumptions used in the NTM to derive these Marginal External Costs are not 100% consistent with the rest of WebTag, but where they differ there is good reason, and the number of these differences and their overall impact is very small.

3.3 The NTM calculates the marginal costs of congestion using a set of speed-flow curves. These are used to represent the relationship between the volume of traffic on a particular link and the speed of the traffic. Congestion is modelled as non-linear. When a link is relatively free of congestion, an additional vehicle will not have a large impact on speed. As the link becomes more congested, an additional vehicle will have a much larger impact upon average speed.

3.4 Within the NTM, congestion is defined as time lost relative to free flow conditions. The speed at free flow conditions is set at the speed limit, adjusted for junctions. As a link becomes congested (and therefore traffic will be travelling at less than free-flow speed) the implied time penalty is modelled.

3.5 The external costs associated with the time penalty firstly consists of the value of journey time increases due to congestion. The NTM combines the modelled delay of a marginal vehicle with the recommended WebTAG (TAG Unit 3.5.6) values of time (VoT) and then sums these across all users of a road to give the cost of delay of an additional vehicle kilometre.

3.6 In addition, the change in vehicle operating costs are taken into account (as in TAG Unit 3.5.6). The addition of a single car will result in a small change in vehicle operating costs caused by a small reduction in average speed for all the vehicles already on the link. Adding these costs to the time costs of delay gives the marginal external congestion costs.

3.7 Estimates of the external costs of accidents, noise, infrastructure damage, local air quality and greenhouse gases (in the form of carbon in carbon dioxide) are calculated in addition to the congestion costs. These are taken from Sansom et al. (2001) with climate change values uprated to reflect subsequent guidance from DEFRA (2002) - such that the cost per tonne of carbon is equal to £70 per tonne in 2000 (2000 prices). Samson et al. (2001) give these marginal external costs by vehicle-type, road-type and area-type for 1998.

3.8 Both NTM results on congestion and the Sansom et al. (2001) estimates of other external costs are for specific years. Interpolation and extrapolation has been used to provide estimates of marginal external costs for the appraisal period. Specifically:

• Values of time have been extrapolated according to TAG Unit 3.5.6. Following the NTM forecasts it is assumed that the proportion of decongestion benefits that accrue to working trips rises from 43% in 2000 to 46% in 2025.
• There is no real growth in infrastructure costs overtime (due to the absence of other evidence).
• Accidents, local air pollution and noise costs are all assumed to grow in line with GDP per capita reflecting increases in people's willingness to pay (as set out in Table 3 of TAG Unit 3.5.6). Air pollution is adjusted for improvements in fuel efficiency (TAG Unit 3.5.6) and tighter emissions standards in line with Defra guidance.
• The cost of carbon is assumed to increase at £1 per year (2000 prices, in line with Defra guidance), but the impact per vehicle kilometre is adjusted for fuel efficiency improvement.

3.9 Indirect tax revenue from fuel duty and VAT on fuel is also estimated based on the average fuel duty of 47.5p in the financial year 2006 staying constant in real terms thereafter. The indirect taxation figures are not likely to be correct before that year when fuel duty was higher in real terms. The NTM models fuel consumption and perceived and resource fuel costs in accordance with TAG Unit 3.5.6. The indirect tax figures presented and taken from NTM and are the difference between perceived and resource fuel costs.

4. Application of Marginal External Costs as Decongestion Benefits

4.1 Table A1 (see Annex A) gives the marginal external costs for cars by area type and road type for 2010 in 2002 prices. A spreadsheet provides more years of data. In the spreadsheet congestion costs are disaggregated by congestion band as defined in Annex B. Road type is based on standard classification of motorways, A roads and other roads. Areas have been classified into three categories - conurbations, other urban and rural - Table A2 (see Annex A) details this classification.

4.2 Identifying the congestion band for a particular road link involves calculating the volume to capacity ratio. Volume is the actual traffic flow; the capacity is the theoretical maximum traffic flow (see Annex B). Both should be measured in the same unit: for example vehicle per time period per road length. However, the specific units would be determined by the data available.

4.3 The next sections outline four steps that can be used to assess the decongestion benefits using the results from the National Transport Model. For rail schemes, specific guidance is available (TAG Unit 3.13.2). A similar approach can be used for a range of schemes which are likely to provide decongestion benefits, such as cycling or walking schemes or investments which improve connections between modes reducing road use.

4.4 Step One: Estimate the change in car kilometres

4.4.1 The first step in applying the marginal congestion costs is to estimate the change in car kilometres due to the scheme. This will be determined by the extent to which car traffic will be diverted off the roads. There will be a diversity of approaches to this assessment depending on the nature of the scheme and its size. The Department should be consulted about approaches which tackle new issues.

4.4.2 In the case of rail, TAG Unit 3.13.2: Guidance on Rail Appraisal: External Costs of Car Use provides some estimates of diversion factors. These have been modelled using the National Transport Model and provide estimates of the change in car vehicle kilometres resulting from a given change in rail passenger kilometres. In the case of cycling and walking interventions, these estimates of diversion factors are also applicable. For more information on the appraisal of cycling and walking schemes, see TAG Unit 3.14.1.

4.4.3 Diversion factors for schemes can also be derived from the experience of previous similar schemes. It will then be possible to provide empirical evidence of the number of car kilometres that have been taken off the road, having suitably adjusted such evidence for the particular circumstances of a scheme. An example of such an evidence base is the evaluation of pilot 'Smarter Choices' (Cairns at al (2004)).

4.4.4 Estimates of the change in car kilometres may also be estimated from a study undertaken specifically for the scheme. A survey of the intention of road users affected by the scheme will quantify the number of journeys that may move from the road so potentially resulting in decongestion benefits.

4.5 Step Two: Analyse the characteristics of the car journeys removed

4.5.1 The results of the NTM marginal external costs analysis (see below) are disaggregated by: road type; area type; and congestion band or level.

4.5.2 In the absence of a highway model, the techniques described below provide for the assignment of the car kilometres estimated in step one to the different road and area types and congestion levels. If feasible and proportionate to the cost of the proposed scheme local evidence should be sought about the routes that would be used. Likely road routes can be identified using highway models or routing software, while traffic flow data for busy roads is available from the relevant highway authority. If possible an opening year estimate and at least one further forecast year estimate should be produced.

4.5.3 Local analysis of the characteristics of the traffic is likely to be most feasible for the opening year estimate. Congestion levels are expected to change over time and routes may also change if, for example, other transport schemes are built. Consideration should be given to how the assignment of traffic might change over time, but this may not be possible in some circumstances. In this case, the same pattern of traffic may be assumed in the future forecast year as the opening year.

4.5.4 Advice from the Department should be sought if it is unclear what effort is proportionate.

4.5.5 In the absence of, or to support, local evidence, estimates of regional traffic flows derived from the NTM can be used. The proportions of traffic in each congestion level for each road type and area type vary by region and these proportions are contained in Spreadsheet 1: Traffic by region, congestion band, area type & road type (MS Excel - 53kb). There are values for 2000 and 2025. Values for any intermediate year can be obtained by linear interpolation. The proportions for 2025 may be assumed if the future forecast year is beyond that date.

4.5.6 In the absence of local evidence, the relevant regional table or tables can be used to with the marginal external costs output from the NTM, which gives estimates per passenger car unit kilometre. The results of the marginal external costs analysis are presented generate a profile of traffic characteristics. If local evidence can provide road and area types but not congestion bands, then the regional traffic proportions can provide evidence on likely congestion bands. For example, if the evidence suggests that a road trip which diverts from rail in the East Midlands will use only rural roads, of which half are trunk & principle and half are 'other', then these two columns of the table for that region can be used to derive the appropriate weights to apply to the diverted car kilometres. The weights show the level of congestion typically encountered by each additional car kilometre in that region for the selected road and area type. If local evidence on the proportions of traffic in each area and road type is unavailable, regional estimates can be found in the statistics; regional data section of the DfT website. Advice should be sought from the Department if the most appropriate method of application is unclear.

4.6 Step Three: Marginal external costs results

4.6.1 Step one and two should provide the change in car kilometres by road and area type and congestion level for the opening year and, usually, at least one other forecast year. These can then be used in Spreadsheet 2: Marginal external costs (MS Excel -1,238).

4.6.2 On the assumptions worksheet the scheme opening year and one other forecast year may be entered. There are options to display the results in different price bases and in pence per mile or kilometre.

4.6.3 The car tables worksheet displays the marginal external costs of car use disaggregated by area type, road type and congestion level. Results are displayed undiscounted for the scheme opening year and the input forecast year. The results change over time as the underlying values of the impacts increase in line with Departmental methodology and factors such as fuel efficiency improve.

4.6.4 The traffic weighed average worksheet takes the marginal external costs of car use as set out in car tables worksheet for the scheme opening year and weighs the congestion costs by congestion band into a single average congestion cost. It also provides a weighed average cost for Great Britain for each separate marginal external cost. The marginal external costs are weighed using GB traffic profiles set out in Spreadsheet 1 (Traffic by congestion band). Depending on the scheme opening year the traffic profile is an interpolation between 2000 and 2025 traffic profiles. After 2025 the 2025 traffic profile is used.

4.6.5 The method described above assumes that the alternative journeys taken in the do-minimum or do-something have the same origin and destination area types. This simplifying assumption is necessary in the absence of a trip distribution model.

4.6.6 The values for each future year can now be combined with the characteristics of the predicted car traffic changes to give the total external costs of those changes for the scheme opening year and the other forecast year.

4.7 Step Four: Calculation of discounted external costs of car use for whole appraisal period

4.7.1 The previous steps will have provided total undiscounted external costs of changes in car use for the scheme opening year and, usually, at least one other forecast year.

4.7.2 Interpolation and extrapolation can be used to derive individual values for all other future years to the end of the appraisal period. Analysts should have regard to the advice in Cost Benefit Analysis (TAG Unit 3.5.4) on extrapolation of benefits.

4.7.3 However, it is recognised that defining reasonable growth profiles for traffic may be difficult for many schemes, particularly those that have used the regional traffic proportions provided above. Factors such as values of time and fuel efficiency should be assumed to continue to grow over time, but in the absence of other evidence road demand (and its allocation to the area and road types/congestion levels) may remain as in the last modelled year. For this unit, this can be done by using the marginal external costs of car use spreadsheet and setting the forecast year to the last year of the appraisal period. The same proportions of traffic used in the last modelled year may then be used to calculate the total change in external costs in the last year of the appraisal period. The profile of road user costs between the last modelled year and the end of the appraisal period may then be estimated by interpolation between the last modelled year and the end of the appraisal period.

4.7.4 These results should then be discounted to the Department's standard base year, which is currently 2002. Cost Benefit Analysis (TAG Unit 3.5.4) also includes advice on discounting.

5. Further Documents

The following documents provide information that follows on directly from the key topics covered in this Unit:

6. References

Cairns, S., L Sloman, C Newson, J Anable, A Kirkbride and P Goodwin (July, 2004) Smarter Choices - Changing the Way We Travel  (available on the Department for Transport website at www.dft.gov.uk/pgr/sustainable/smarterchoices/ctwwt/).

Clarkson, R., Deyes, K. (2002) Estimating the Social Costs of Carbon Emissions', Department for the Environment,Food and Rural Affairs, London. Can be found at
www.hm-treasury.gov.uk/documents/taxation_work_and_welfare/taxation_and_the_environment/tax_env_GESWP140.cfm.

Sansom, T., Nash, C., Mackie, P., Shires, J., & Watkiss, P. (2001) 'Surface Transport Costs & Charges: Great Britain 1998' Department of the Environment, Transport and the Regions, London.

7. Document Provenance

This Transport Analysis Guidance (TAG) Unit is a revision of the April 2004 version.  It provides new guidance on the use of estimates of the marginal external costs of adding car traffic to a road in the estimation of decongestion benefits.

Technical queries and comments on this 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

8.1 Annex A: Marginal External Costs and Area Type Classification

Table A1: Marginal external costs for cars (p/km, 2010, 2002 prices)

Table A2: Area type classification

8.2 Annex B: Congestion Band Definition

The congestion bands used in the external costs spreadsheet reflect the volume to capacity ratio of a traffic link. The volume (v) is the actual traffic flow and the capacity (c) is the theoretic maximum traffic flow. These can be expressed in terms of vehicle (or PCU (passenger car unit)) per time period per road (or lane) length. Table 7 shows how the congestion bands relate to the ratios.

Table B1: Congestion bands in terms of volume over capacity

When assigning traffic to the v/c bands the process assumes "average network" lane capacities. However, depending on local conditions, the actual capacity of a link may be somewhat more or less than the capacity assumed at the site. In some cases actual flows may exceed the theoretical capacity of a link and lead to v/c ratios in excess of 1.

Appraisals should seek to identify the capacities of roads that are used as substitutes for rail, if possible and proportionate for the size of the scheme. In the absence of more local knowledge, tables 8 and 9 contain suggested capacities for roads in rural and urban areas respectively. Table 10 shows the PCU factors for different vehicle types.

Table B2: Suggested average capacities (PCU per lane km per hour) for rural roads

Table B3: Suggested average capacities (PCU per lane km per hour) for urban roads

Table B4: PCU factors by vehicle type