Guidance on Rail Appraisal: External Costs of Car Use
TAG Unit 3.13.2

April 2007

Contents

1. External costs of car use
1.1 Introduction

2. Application of marginal external costs
   2.1 Elasticity based rail appraisal
   2.2 Step one - estimate the change in car kilometres
   2.3 Step two - analyse the characteristics of the car journeys
   2.4 Step three - marginal external costs results
   2.5 Step four - calculation of discounted external costs of car use for whole appraisal period

3. Presentation of results
   3.1 Appraisal tables
   3.2 Spreadsheet of results

4. Further information

5. References

6. Document provenance

7. Annex A - Definitions
   7.1 Road and area types
   7.2 Congestion bands

1. External costs of car use

1.1 Introduction

1.1.1 This TAG unit explains a method for assessing the external costs^[1] of road use in a rail appraisal. The use of road vehicles incurs private costs (such as fuel costs and personal travel time) and external costs. These externalities are costs imposed by an action that are not experienced by the individual who carries out the action. For car use these costs include congestion, air pollution, noise, infrastructure costs and some accident costs. Whilst this is not a complete list of the external consequences of road traffic, DfT is in a position to estimate monetised impacts for these categories and therefore the guidance focuses on these^[2].

1.1.2 Most rail schemes have an effect on the use of other modes which will result in a change in the external costs of car use. A multi-modal model can provide robust estimates of these external costs. However, for a number of reasons, including the relatively low share of rail in the market for travel, elasticity based modelling of rail demand has been the established practice for schemes other than major improvements in urban areas. Elasticity based models are not capable of providing direct estimates of these external costs as the road network is not represented.

1.1.3 Multi-modal models are the best way of identifying the impact of changes in one mode on all other modes. Where such models have been set up with the intention of informing rail appraisal, they should be used to provide estimates of road decongestion benefits. In areas where multi-modal or highway models exist which were not set up for the explicit purpose of rail modelling and appraisal, rail scheme sponsors should consider whether these models can be used to provide estimates of the decongestion benefits. Guidance on possible approaches is given in Major Schemes Appraisal: Road Decongestion Benefits (TAG Unit 3.9.5) The Department can advise on the models that exist, the areas that they cover and on how best to derive the required estimates of cross modal effects. 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.

1.1.4 However the Department understands that, due to the potentially high costs of multi-modal studies and/or lack of access to the appropriate tool, this may not always be possible. Therefore the following guidance can be used provided at least one of the following criteria is satisfied:

• A capital expenditure of less than £25m

Or,

• The business case for the scheme is not sensitive to the value of the benefits derived from a change in the external costs of car use. This can be tested by removing 50% of the benefits as calculated using this guidance. If the BCR falls below 1.5 then further analysis of the external costs of car use will be required.

1.1.5 Where neither of these conditions is satisfied reference should be made to the methods set out in TAG Unit 3.9.5. Guidance on highway network modelling is available in the Design Manual for Roads and Bridges (DMRB) Section 12.2.1, while guidance on variable demand modelling is available in TAG Unit 3.10 and TAG Unit 3.11. You may wish to contact the Department to discuss whether a more appropriate modelling tool is available for use in appraising the scheme.

1.1.6 Marginal external costs of car use have been estimated from the Department's National Transport Model and Surface Transport Costs and Charges: Great Britain 1998 (Sansom, Nash, Mackie, Shires & Watkiss, 2001). Derivation of these costs and generic advice on their use will be described in a separate TAG Unit, currently in preparation. This unit explains a method for using these in elasticity based rail appraisals.

1.1.7 The approach to estimating the external costs set out in this TAG Unit provides an approximation to be used in conjunction with elasticity based models for rail. The estimates have been derived from the National Transport Model (see the DfT website for more information). They do not take into account all of the responses available to those who switch mode or the effect of the initial change in traffic levels on costs and subsequent demand. Sensitivity testing of scheme appraisals to the results of this approach will therefore be expected.

2. Application of marginal external costs

2.1 Elasticity-based rail appraisal

2.1.1 The method described here is applicable to rail schemes that have used elasticity-based modelling, such as that found in the Passenger Demand Forecasting Handbook. However, it should only be used in the circumstances described in paragraph 1.1.3.

2.1.2 Elasticity-based modelling of a rail scheme will need to provide changes in rail passenger kilometres between stations on the rail network for the scheme opening year and, usually, at least one other forecast year^[3]. Several steps need to be taken to estimate the change in the external costs of car use from this information. Steps one to three calculate total changes in external costs for the opening year and the future forecast year, and then step four explains how this analysis can be extended to cover the whole appraisal period.

2.2 Step one - estimate the change in car kilometres

2.2.1 Where rail passenger kilometres change there are likely to be changes in total kilometres travelled and in kilometres travelled by other modes. In order to estimate the change in external costs of car use, the change in car kilometres needs to be calculated. Where possible, the change in car kilometres should be estimated using local evidence such as passenger surveys.

2.2.2 In the absence of local evidence, diversion factors based on the National Transport Model (NTM) may be used to convert a change in rail passenger kilometres to a change in car kilometres. The following paragraphs outline the method used to estimate the national average diversion factors given in Table 1. Although these diversion factors are based on an assumed change in rail fares, they are considered to provide some evidence for most measures leading to diversion to, or from, rail. However, it is recognised that this method does not reflect local factors and therefore sensitivity tests around the assumptions used will need to be provided.

2.2.3 Furthermore, for some schemes these factors will not be applicable, and scheme specific evidence based on local studies will need to be supplied. This would include schemes where long distance access trips by car are likely to be affected or where the purpose of a scheme is to encourage mode shift. All scheme appraisals will need to consider whether the nature of the scheme is likely to make the national factors inappropriate.

2.2.4 Two scenarios in the NTM were compared to derive these factors. The base case has rail fares remaining at 2003 levels until 2025; this is compared to a scenario which introduces RPI+1 per cent per annum increase in fares over the same time period. The results are presented in table 1 as a percentage of the change in rail kilometres.

2.2.5 A fall in rail passenger kilometres is observed in the model due to the relative increase in rail costs, and hence an increase in the use of other modes. Overall distance travelled is estimated to fall by 46% of the fall in rail kilometres as the trips replacing rail trips are generally shorter distance. 26% of the decrease in rail passenger kilometres is replaced by car driver kilometres. The information on the mode share of distance travelled in table 1 is required for estimating the external costs of the rail related change in road use. The use of these factors converts a change in rail passenger kilometres to a change in car kilometres, taking into account car occupancy rates.

Table 1: National average diversion factors from the National Transport Model - Changes in distance travelled as a percentage of a change in rail passenger kilometres

For the purposes of this unit, the 20% change in car passenger kilometres does not affect the change in car kilometres. All values are quoted to 2 significant figures.

2.2.6 The mode share of all trips in table 2 is shown for the purpose of illustrating the outputs from the NTM but is not required for estimating the change in car kilometres. In the NTM the total number of trips does not change with a change in costs, hence the total change in trips in table 2 is zero. However, length of journey and the destination area type choice are possible responses to changing costs.

Table 2: National average diversion factors from the National Transport Model - Changes in trips as a percentage of a change in rail trips

2.3 Step two - analyse the characteristics of the car journeys

2.3.1 The results of the NTM marginal external costs analysis (see below) are disaggregated by:

1. road type;
2. area type; and
3. congestion level.

Definitions of the above are included in Annex A.

2.3.2 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 between rail stations 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.

2.3.3 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.

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

2.3.5 In the absence of, or to support, local evidence, estimates of regional traffic flows can be derived from the NTM. The proportions of traffic in each congestion level for each road type and area type vary by region. These proportions are contained in Spreadsheet 1. 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.

• Spreadsheet 1: Traffic by region, congestion band, area type & road type (MS Excel - 53kb)

2.3.6 In the absence of local evidence, the relevant regional table or tables can be used to give a profile of traffic characteristics. If local evidence can provide road and area types but not congestion bands, then the regional traffic tables 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. These weights will indicate the level of congestion typically encountered by each additional car kilometre in that region for the selected road and area type.. Advice should be sought from the Department if the most appropriate method of application is unclear.

2.4 Step three - marginal external costs results

2.4.1 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 with the marginal external costs output from the NTM, which gives estimates per passenger car unit kilometre. The indirect taxation external cost information in the spreadsheet should be ignored, instead refer to section 3.7 "Indirect Costs to Government" of TAG 3.13.1 for guidance on calculating indirect tax changes. The results of the marginal external costs analysis are presented in this spreadsheet:

• Spreadsheet 2: Marginal external costs (MS Excel - 1,238kb)

2.4.2 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.

2.4.3 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.

2.4.4 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.

2.4.5 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.

2.4.6 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.

2.5 Step four - calculation of discounted external costs of car use for whole appraisal period

2.5.1 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.

2.5.2 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.

2.5.3 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.

2.5.4 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.

3. Presentation of results

3.1 Appraisal tables

3.1.1 DfT standard rail appraisal tables should be used for all appraisals, see TAG Unit 3.13.1. Results from the methodology explained in this unit should be included in these tables as follows:

1. The estimated change in congestion costs should be entered in the Transport Economic Efficiency table as a change in consumer travel time for cars, LGVs and goods vehicles. It should be noted that the calculation of 'congestion' cost includes an estimate of vehicle operating cost changes.
2. The estimated changes in greenhouse gases, local air quality, noise and accident costs should be entered in the relevant boxes of the Analysis of Monetised Costs and Benefits table.
3. Road related infrastructure costs will generally accrue to the Highways Agency or Local Government and should therefore appear in the Public Accounts table under the central or local government investment costs headings.
4. A note should be added to all tables to explain that the methodology in this unit has been applied.

3.1.2 All values estimated in this unit should also be included on the Appraisal Summary Table with a note to explain how they were estimated.

3.2 Spreadsheet of results

3.2.1 It is anticipated that the above method will require the use of spreadsheet software to calculate the total external cost change estimate. A clear spreadsheet of all calculations, assumptions and results must be submitted with any scheme that uses this methodology. The separate totals for each category of benefit calculated using this methodology (e.g. congestion, greenhouse gases, etc.) should be stated clearly in scheme documentation.

4. Further information

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

5. References

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.

6. Document provenance

This Transport Analysis Guidance (TAG) Unit provides new guidance on the topic.

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

Appraisal Team - Rail Network Analysis and Modelling
Department for Transport
Zone 4/33 Great Minster House
76 Marsham Street
London, SW1P 4DR
E-mail: railtag@dft.gsi.gov.uk
Tel: 020 7944 6751
Fax: 020 7944 2160

7. Annex A - Definitions

7.1 Road and area types

7.1.1 This Annex contains information and tables explaining the definitions of terms used in FORGE (Fitting On of Regional Growth and Elasticities) and the outputs provided in this unit.

7.1.2 Table 3 shows the codes assigned to different area and road types that the results spreadsheet uses. All motorways outside conurbations are assumed to be in rural areas for the purposes of the model.

Table 3: Specification of Conurbations, Other Urban, Rural, Motorways, A roads and B&C roads in terms of FORGE area and road type codes

7.1.3 Table 4 shows the descriptions of the different FORGE area type codes.

Table 4: FORGE area type codes

7.1.4 Table 5 shows a detailed definition of the FORGE area types.

Table 5: FORGE area types

7.1.5 Table 6 gives a description of the FORGE road type codes.

Table 6: FORGE road codes

7.2 Congestion bands

7.2.1 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 7: Congestion Bands in terms of volume over capacity

7.2.2 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.

7.2.3 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 8: Suggested average capacities (PCU per lane km per hour) for rural roads

Table 9: Suggested average capacities (PCU per lane km per hour) for urban roads

Table 10: PCU Factors by Vehicle Type