TAG Unit 3.15.1: Forecasting Using Transport Models

July 2008

1. Forecasting Using Transport Models
    1.1 Introduction
    1.2 The Principles of Forecasting
    1.3 Model Design Considerations
    1.4 Development of the Reference Case
    1.5 Forecast Scenarios: The Without-Intervention and With-Intervention Cases
    1.6 Reporting Requirements

2. Further Information

3. References

4. Document Provenance

1. Forecasting Using Transport Models

1.1 Introduction

1.1.1. Forecasting the usage (traffic flows, public transport patronage and so on) and performance (journey times, crowding and so on) of transport networks in the future is a critical component in any transport appraisal. The impact of a policy or intervention may only be assessed satisfactorily with a robust forecasting methodology. This TAG Unit provides guidance for the forecasting of all modes of land-based transport. This incorporates advice across the whole spectrum of appraisal: for those wishing to appraise local transport interventions, to regional and national policy-makers.

1.1.2. Transport appraisals are only as good as the forecasts that inform them. It is important to produce accurate forecasts for the scope of the appraised intervention. The need for credible forecasts of traffic in future years is of fundamental importance in appraisals, where it is important to know, and present to the public, the impact the intervention will have when it is implemented. There are also the interventions where conditions on the network can be expected to change over the foreseeable future, and the implications for the economic and environmental assessments must be considered.

This TAG Unit

1.1.3. The general discussion in this TAG Unit gives practical guidance to undertaking the forecasting, using transport models, of transport for policy options in the future. This includes issues such as how to structure a transport model and obtaining the data required. This builds on the standard forecasting framework recommended by the Department. TAG Unit 3.1.2 Section 5 lays out the main steps in the forecasting process and the rest of the Unit gives a useful overview of transport models in general.

1.1.4. This Unit ties together the other material in this family of TAG Units relating to forecasting, including the following:

• The principles of forecasting in Transport Models (TAG Unit 3.1.2)
• Use of TEMPRO Data (TAG Unit 3.15.2)
• Forecasting and Sensitivity Tests for Public Transport Schemes (TAG Unit 3.15.3)
• Rail Passenger Demand Forecasting Methodology (TAG Unit 3.15.4)
• The Treatment of Uncertainty in Model Forecasting (TAG Unit 3.15.5)
• Generalised cost formulation, values of time and detailed data sources for the development of the reference case (see VDM - Scope of the Model, TAG Unit 3.10.2)

1.1.5. This Unit does not explore in any depth the potentially complex interaction between land use and transport and the forecasting of land use growth outside that provided through TEMPRO. For more information on this issue see Land-Use / Transport Interaction Models (TAG Unit 3.1.3).

1.1.6. This Unit does not provide guidance on wider forecasting topics. In particular, it does not give guidance on the use of environmental forecasting models (for noise and air quality, for example - see the sub-objectives in The Environment Objective, TAG Unit 3.3), nor on forecasting future economic parameters (GDP growth, value of time, vehicle operating costs and so on - see Values of Time and Operating Costs, TAG Unit 3.5.6).

1.2. The Principles of Forecasting

1.2.1. This section outlines the general principles of forecasting in transport models. This has been taken from section 2.12 of Transport Models (TAG Unit 3.1.2).

1.2.2. The main aim of forecasting in transport models is to determine how patterns of demand will shift over time, reflecting exogenous factors such as growth in incomes, changes in transport prices and demographic and land-use changes. These matters will now be discussed.

1.2.3. A basic procedure underlying the vast majority of practical transport models is a 'trip generation' stage, which relates the general volume of travel (separately, in most cases, for a number of distinct journey purposes) to person-type characteristics. Regardless of how these relationships are derived, they tend to be applied at the zonal level: in this way they are sensitive to the (forecast) numbers of persons of different types in the zone.

1.2.4. As a minimum, zonal populations and employment will be required, together with the number of households, usually broken down by level of car ownership. Some breakdown by age and employment status is desirable, as is the availability of driving licences. The level of detail should be at least as much as that required for the implementation of the demand model. Thus, if it is decided that the demand model should distinguish between different levels of income, then the model will require forecasts of the population at these different levels.

1.2.5. 'The trip generation process provides inputs for the demand model (see Variable Demand Modelling - Scope of the Model, TAG Unit 3.10.2). However, as a general guidance, the number of distinctions made within the demand model tends to be low. Since models of trip generation are relatively easy to develop, there is a case for allowing for a greater level of 'segmentation'. At the same time, the DfT has developed forecasts at the 'NTEM zone' level, a disaggregation of the local authority level that is used in the National Trip End Model. Unless there are good reasons for doing otherwise, the Department recommends that the modeller should make use of these. Since the zones in the study area will typically be more detailed than NTEM zones, some method of spatial disaggregation will be required. To ensure national consistency and to avoid optimism bias, it is required to control the predicted growth to the DfT forecasts, at least at the modelled area or study area level.

1.2.6. The predictions about future growth in demand may be expressed in the form of a forecast year demand curve, as illustrated in Figure 1. The demand curve illustrates the total demand for travel given a cost per trip. Consider a base year description of travel movements (set of matrices) assumed to be in equilibrium at point A. At this point, the average cost of a trip, CA, results in the volume of trips TA being made. In future years, socio-demographic changes such as increases in population and car ownership cause the demand for travel to increase. This is represented by a shift in the demand curve; ie at the same level of cost, more trips are demanded to be made. This is shown by point B and constitutes the Reference Case position. Costs remain at CA, whereas trips made increases to TB. Hence, it can be seen as a way of locating the future demand curve, assuming a constant functional form.

Figure 1: The shift in the demand curve over time

1.2.7. It is important to note that this 'reference case' is not intended to represent a realistic forecast of what might happen - i.e. it is not an equilibrium solution. Typically, the increased growth will lead to cost changes (e.g. greater congestion) through the supply relationships.

1.2.8. Having located the future demand curve, the transport model will be run in order to balance the supply and demand and converge to an equilibrium position. This is represented by point C. Note that, in addition, there may be forecast changes in supply (essentially the distinction between 'without-intervention' and 'withintervention' cases).

1.2.9. In this context, cost forecasts relating to, for example, the price of petrol or public transport fares, can be viewed as changes in the supply functions. At equilibrium, the users of the transport network have minimised their generalised cost (a weighted combination of time and money), in which units the demand curves are specified. To reach this point requires consideration of how users value their time.

1.2.10. Generalised cost is an underpinning concept in transport models. It is the generalised cost that the traveller attempts to minimise to travel from their origin to destination. This will vary according to the individual, and is represented in transport models by segmentation into user classes, defined by incomes, car ownership and so on, and different journey purposes, such as work and non-work journeys. Each of these different categories and journey purposes will have different values of time and hence sensitivity to the cost of travel.

1.2.11. More detail on updating the generalised costs and values of time for forecast years can be seen in Section 1.5.

1.3. Model Design Considerations

1.3.1. Figure 2 shows the basic structure of a transport model with reference to what is required in forecasting. The forecasting process begins with the development of a reference case by updating demand factors to each forecast year being appraised. This effectively locates a point on the demand curve for those future years. Supply-side factors are then updated (ie network changes, different cost assumptions) to derive the most likely 'without-intervention' scenario against which one may test the impact of various schemes and policies that are to be introduced in the 'with-intervention' scenarios.

Figure 2: Basic approach to forecasting using a transport model

Scope

1.3.2. The design of the models required to produce forecasts should depend on the scope and scale of the interventions to be appraised. For example, small scale urban traffic models will be treated differently from more complex multi-modal models at the regional level.

1.3.3. The recommended approach is to adopt methods most suitable to the interventions to be tested. A simple fixed trip matrix approach is appropriate where it is judged that the transport network will not become stressed with congestion in the forecast years. This assumes that there will be no induction or suppression in the demand for trips in any of the future years tested. Where the model must be able to adequately account for responses to congestion, a variable demand model must be used. Where the scope of the model is uncertain, the model may be tested after the preparation of the reference case when assessing the potential for congestion. Guidance on assessing model scope and conducting testing on the reference case to ascertain this is in Variable Demand Modelling - Preliminary Assessment Procedures (TAG Unit 3.10.1).

1.3.4. Detailed advice on the whole variable demand modelling approach is described in detail in VDM - Detailed Stages (TAG Unit 3.10).

1.3.5. Irrespective of the use of a fixed trip or variable demand model, the analyst will have in hand a validated base year trip matrix. This provides, in the base year for each origin zone, the total number of trips. As a minimum, any multi-stage demand model will require a database which provides, in the base year for each origin zone, the total number of trips:

• to each destination zone,
• for each of several purposes,
• by each of the modes modelled, and
• within each modelled time period.

Selection of Forecast Years

1.3.6. The selection of forecast years is an important consideration in model design. This decision largely depends on the realistic time by which individual interventions to be tested are likely to be implemented. This decision may also be influenced by the availability of forecast data and the requirement of how far into the future the appraisal is preferred to be undertaken. Details concerning base and forecast year selection for formal transport models and projecting these forecasts forwards to fulfil the appraisal period are in Cost Benefit Analysis (TAG Unit 3.5.4).

1.4. Development of the Reference Case

1.4.1. Having established a base year case in the transport model, the next step is to develop a reference case based on demographic changes as a basis for the estimation of future demand (as described in Section 1.2). It is from the reference case that the forecast cases are developed by changing the supply conditions (network changes, fuel prices, public transport fares, etc). A new equilibrium point is then established between supply and demand. This is elaborated in the next section.

Reference Case Data Requirements

1.4.2. Forecasts of transport demand must be based on the validated trip matrices developed for the base year, with separate forecasts being made for each model time period.

1.4.3. The principle source of data to produce a reference case from the base year is available from TEMPRO (see Use of TEMPRO (TAG Unit 3.15.2)). This provides growth factors to apply to the transport model in hand. This data set provides forecasts of growth in trip ends and planning data for any range of years between 1991 and 2041. Reference cases need to be prepared for each forecast year that is required in the model running.

1.4.4. Forecasts should be controlled to the benchmark provided by the TEMPRO data. This is because different schemes compete for a common budget, which should be planned on the basis of common assumptions about growth. Using centrallycontrolled forecasts allows the results of models to be directly comparable to those elsewhere, reducing the likelihood of local optimism bias.

1.4.5. All of the growth factors necessary to satisfy these requirements are available from the TEMPRO software. This is freely available from the TEMPRO web site, www.tempro.org.uk.

1.4.6. The level of segmentation within the transport model will determine the level of detail at which data must be collected and input. More complex strategic scale models are often more highly segmented than their urban-level counterparts that have more spatial detail and less user class groups. For more guidance of the level of segmentation appropriate for the model and a checklist of recommended additional data sources, see VDM - Scope of the Model (TAG Unit 3.10.2).

1.4.7. The following planning data growth factors are available from TEMPRO:

• Population by age (0-15, 16-64, 65+)
• Households
• Workers (employees at the home end)
• Jobs (employees at the work end)

1.4.8. The use of local land use forecasts will add accuracy to models when looking at a more detailed level (ie below the NTEM zone level). As mentioned, these should be controlled to TEMPRO totals at a higher spatial level. It must be recognised when altering local land use information that uncertainty exists as to whether developments will or will not go ahead, as well as causal linkages between developments. Sensitivity tests should be used to test the impact of the assumptions on the robustness of the model. See Treatment of Uncertainty in Model Forecasting (TAG Unit 3.15.5).

Forecasting Growth in Highway Demand

1.4.9. Growth rates for trip matrices are primarily derived from national growth in trip ends from TEMPRO (origins and destinations of trips) - see Use of TEMPRO Data (TAG Unit 3.15.2). TEMPRO provides growth in trip matrices for different modes, purposes, time periods and car ownership categories.

1.4.10. Where estimates of traffic growth are required, for example for more simple models or where no model is available, national traffic growth forecasts from the National Transport Model can be usefully combined with local trip growth from TEMPRO. Guidance on this can also be found in TAG Unit 3.15.2.

1.4.11. It is also acceptable to use growth rates taken from a higher-tier model where available, especially in the case of models of urban areas. This in principle will account for the majority of the redistribution and mode shift effects resulting from differential land use changes and the influence of major transport interventions. The higher tier models will therefore be able to supply more local models with more detailed growth factors. The higher tier model will have been thoroughly validated and will be in accordance with the NTEM growth factors at that level, hence maintaining the national consistency within the local model.

Forecasting Growth in Public Transport Demand

1.4.12. Reference case estimations of heavy rail growth are available from TEMPRO. Alternative forecasts are possible by using the elasticity-based methodology recommended in Methodology for Forecasting Rail Demand (TAG Unit 3.15.4). This approach builds on the method set out in the revised Passenger Demand Forecasting Handbook (PDFH) (ATOC, 2005). This method forecasts by putting a greater emphasis on growth in GDP.

1.4.13. It is recommended that for heavy rail demand forecasting using elasticity-based methodology, TEMPRO planning data control totals are used. This ensures consistency between other rail schemes as well as schemes for other modes. The TEMPRO planning data controls the population, employment, etc. that will have a significant impact on future demand levels.

1.4.14. Forecasting of other public transport modes to any level of sophistication requires a multi-modal transport model. For the reference case, the analyst is required to use forecasts of future demand for each mode (normally bus and light rail) to growth up the public transport matrices. These forecasts are available from TEMPRO and may be applied in the same way as for private vehicles.

Forecasting Growth in Freight Traffic

1.4.15. Growth factors to be applied to freight matrices are not available from TEMPRO. A useful source is the regional traffic forecasts published by the Department. This source provides expected growth rates of traffic by vehicle type and area type in each Region. Presently these are available from 2003 to 2025. Beyond this period, these forecasts should be extrapolated to the required modelled year.

Forthcoming guidance on freight modelling shall elaborate on this. If more guidance is required, contact ITEA Division.

1.5. Forecast Scenarios: The Without-Intervention and With-Intervention Cases

1.5.1. At this point, we have determined the level of demand in the reference case and now it is a matter of adjusting the supply conditions for the future year. This next step is to develop a realistic future year case, ie the "without-intervention" scenario. This constitutes the core scenario in the future against which one may compare results from "with-intervention" policy tests to appraise the impact of the interventions that are required to be tested.

1.5.2. This section describes the development of a forecast year in general, getting from the reference case to a realistic representation of conditions of supply of and demand for transport. In practise the analyst will develop the without-intervention case first and then augment changes that are required in the with-intervention cases. In principle, the data requirements are the same. At the end of this section a summary is given of how the without-intervention and with-intervention cases should be compared.

Requirements for a Forecast Year

1.5.3. The core without-intervention case must also include the most likely development of land use and transport interventions that influence the transport network. These can be derived from local plans. Forecasting further into the future will require consultation with the planning bodies to ensure acceptable forecasts. Alternative without-intervention scenarios may also be required, to reflect uncertainty about the evolution of future plans. For more detail on how to assess a most likely scenario and resulting issues of robustness to uncertainty in alternative scenario development, see The Treatment of Uncertainty in Model Forecasting (TAG Unit 3.15.5).

1.5.4. The with-intervention cases must be based on the without-intervention scenarios but will also include those interventions that are to be appraised to allow decisionmakers an informed view of the impact and value for money of different options.

1.5.5. Changes to the transport network will also affect travel times and journey costs. For example:

• Building of new roads or railways
• Alteration of existing infrastructure (e.g. road widening)
• Changes to public transport routes or level of service (e.g. headways)
• Introducing or changing pricing structures for road, public transport or parking
• Additional or removal of public transport nodes (e.g. rail stations and bus stops)
• Soft engineering measures such as changing speed limits, traffic calming, junction redesign, etc.
• Volume of parking spaces available

Generalised Costs

1.5.6. Generalised cost is the measure of disutility of a journey from origin to destination across the transport network. This is a combination of monetary cost and journey time (in some cases, generalised cost may also include terms reflecting quality, such as crowding or reliability). Guidance in line with best practice would be to assume that generalised cost is defined in units of time, on the principle that time is a stable unit of measurement. Money becomes 'less valuable' over time with increasing incomes and requires the model to account for this.

1.5.7. Where a demand model is being used, generalised cost is an essential component, responding to network costs received from the supply model in order to estimate elements such as trip distribution and mode choice. The use of generalised cost is also strongly advised in the assignment phase (supply), where methods such as time-only assignment should be reserved for model development purposes and potential sensitivity tests, or where such an approach may be rigorously justified as being satisfactory for the scope of the project.

1.5.8. More detail on the concept of generalised cost and its formulation is given in Section 1.10 of VDM - Scope of the Model (TAG Unit 3.10.2). The recommended growth in factors pertaining to generalised cost for each traveller segment is given in Values of Time and Operating Costs (TAG Unit 3.5.6). These must be used to update the base generalised cost assumptions to those in forecast years and must be the same in the without-intervention and with-intervention cases in order to allow direct comparability.

1.5.9. The main future year changes to costs that need to be incorporated into the model are as follows. These are subsequently explained in further detail:

• Values of time (VOT). This governs the relative generalised cost of a journey based on the type of person travelling.
• Vehicle occupancy. This is used to account for passengers in the calculation of generalised costs for the average vehicle.
• Vehicle operating costs (ie fuel prices, vehicle efficiency and vehicle occupancy). Standard growth rates should be applied to growth up these factors.
• Background economic factors (GDP, incomes and car ownership levels). See TAG Unit 3.5.6 for general macroeconomic variables and Use of TEMPRO (TAG Unit 3.15.2) for car ownership levels. Note that TEMPRO tripend forecast data takes account of GDP and income growth.
• Public transport fares. These are largely determined by general inflationary increases or from specific changes planned by operators. See Forecasting and Sensitivity Tests for Major Public Transport Schemes (TAG Unit 3.15.3).
• Tolls and road user charges. These are determined by specific planned interventions. These affect the generalised cost of a journey and hence different types of people will have a different propensity to pay, hence affecting route and mode choice.
• Parking charges. These are set by the local authorities. These also influence the generalised cost of a journey.

1.5.10. It is important that the values of time, vehicle operating costs and vehicle occupancies are appropriate for the forecast year being modelled. TAG Unit 3.5.6 demonstrates how these figures are derived and how base year values may be factored to the forecast years that are required.

1.5.11. The value of time is generally assumed to increase as income increases. It is also usually assumed that values of time should increase over time in line with the forecast growth in GDP. TAG Unit 3.5.6 details how the value of time is expected to change over time for use in appraisal.

1.5.12. For modelling purposes, these growth projections should also represent a reasonable expectation of future values. The values of time represent the travellers' perceptions of the time they have to spend travelling, and their willingness to trade money for time in order to visit their destinations. This is usually different from the standard base year values used in appraisal. Base values of time used in transport models may be derived by making use of more detailed local data where available (such as stated preference surveys) and justified through adequate model calibration and validation. In principle there should be different VOT distributions for each type of user class - for different journey purposes, for cars and goods vehicles, for travellers in different SEG or income groups, and for peak and off-peak travel. VOT has also been found to increase with trip distance, and the inclusion of this relationship may need to be considered where there is a wide distribution of trip lengths within (say) a transport corridor.

1.5.13. Different types of vehicle also have different values of time that must be accounted for in the forecast years. This concerns different classes of freight vehicles, for which forecast growth over time varies between types. The relative proportions of freight vehicles are accounted for in the preparation of the reference case, taking growth rates from NTM between the years required in the model. From this the value of time for an average vehicle may be determined on a particular route.

1.5.14. The changes in value of time should be reflected in a demand model where applicable. This ensures the validity of demand responses to changes in generalised cost parameters in the future.

1.5.15. In the case of assignment models, it is recommended that the value of time should also change over time. This should be in line with the changes in the demand model where one is used. Changing values of time for modelled user groups in the assignment stage will impact on route choice decisions. Some policy testing may be particularly sensitive to this, such as tolling or road pricing policies. Fixed values of time in assignment models should only be used where the impact is considered to be negligible given the scope of the model and some evidence should be provided to support this assumption.

1.5.16. See VDM Scope of the Model (TAG Unit 3.10.2) for further discussion on traveller values of time.

1.5.17. Vehicle occupancy is used to account for passengers in the formulation of generalised cost in the assignment process. Values of time for one person for a certain purpose and vehicle type are converted into an average car value of time in order to calculate the generalised cost for those travelling. Therefore vehicle occupancy needs to be combined with the value of time in order to do this. This information is available from TAG Unit 3.5.6.

1.5.18. Vehicle operating costs for private vehicles include factors such as changes in fuel prices and future fuel efficiency levels. Projections of both are available in TAG Unit 3.5.6. For public transport, rather than vehicle operating costs behaviour will be determined by fares and other factors, such as image and overcrowding. These may change the generalised cost of travel on such modes, although they may often be attributed on a service-by-service basis.

1.5.19. Forecasts of changes in public transport fares can be problematic, especially over longer forecasting periods. For more information, see Forecasting and Sensitivity Tests for Public Transport Schemes (TAG Unit 3.15.3). For more information on forecasting the generalised cost for rail specifically, see TAG Unit 3.15.4.

1.5.20. In the case of road pricing, a similar approach to that taken for parking charges may be appropriate. There may be real changes in the costs of operating the scheme which it may be decided should be reflected in a real change in the prices. It is much more likely that real increases in the prices are required in order for the scheme to continue to meet its congestion reduction target. Again, some experimental testing may be required to determine the appropriate level. Account should also be taken of any stated policies about road pricing over time.

1.5.21. Changes in road tolls may be simple to deal with in some cases and less so in others. In some instances, the tolls are controlled by Government and the policy about future increases is clear. In other cases, such as the M6 Toll Road, the toll levels are at the discretion of the operator. In the absence of clear policies, a sensible approach may be to assume no real change and, if potentially important to the case for the public transport scheme, the effects of changes in toll levels may be assessed by means of sensitivity tests.

1.5.22. Parking charges may be expected to change in real terms for two main reasons:

• the costs of operating car parks may change at a rate different to the rate of inflation; and
• charges may be changed in order to change demand for parking.

1.5.23. The effect of the first reason may be sufficiently small to ignore. In contrast, the effect of the second reason could be appreciable. It is important to determine local parking strategies in order to incorporate these policies into the model appropriately. This is particularly important where it may be difficult to assess parking charges aimed at controlling demand over a long forecasting horizon.

1.5.24. Only where a specific local policy measure of parking demand management through charging exists, a possible approach may be to assume just sufficient increase in charges for the demand for parking to fit within the supply. It may be necessary to run the model several times in order to find the appropriate parking charges that will reduce demand to the level that can be accommodated. Where there is a clear policy on the level of parking charges, and the resulting charge would reduce demand to below the supply, the charges resulting from the policy may be assumed.

1.5.25. Where a demand-management pricing policy is not in place, for zones where demand for parking exceeds supply, alternative means of demand restriction will be necessary in the model. These may include, for example, increasing egress times or increasing the disutility for parking in affected zones.

Supply Factors

1.5.26. Supply-side changes include physical changes to the network, changes to public transport services and all other relevant changes to which the demand for travel will respond.

1.5.27. In many models, parking supply is not represented explicitly, but difficulty in parking in central areas is represented by both a money cost and a non-zero parking time associated with the centroid connector for the car mode. These money and time penalties may have a significant impact on future year mode choice, and the future year values need to be considered carefully.

1.5.28. Further guidance on forecasting the future supply of public transport services is in Forecasting and Sensitivity Tests for Public Transport Schemes (TAG Unit 3.15.3).

The With-Intervention Cases

1.5.29. The only difference between the without- and with-intervention cases is the presence of the intervention itself.

1.5.30. A fixed trip approach treats trip matrices as constant between scenarios. A variable demand approach allows demand responses to occur given a change in transport conditions (see TAG Unit family 3.10). It is strongly recommended that variable demand modelling is used in preference to a fixed trip approach where this scope of intervention(s) is sufficient (see section 1.2 in VDM - Preliminary Assessment Procedures (TAG Unit 3.10.1)). In addition, Methodology for Forecasting Rail Demand (TAG Unit 3.15.4) gives guidance on how rail patronage may be affected by interventions introduced as well as how traffic may respond to rail interventions.

Accounting for Uncertainty

1.5.31. To account for uncertainty, sensitivity test cases for the without-intervention and with-intervention scenarios should be established in order to derive a range to cover the effects of uncertainty in the results. This is to account for the broad spectrum of uncertain elements in a forecast, especially for the appraisal of potential interventions in the more distant future.

1.5.32. Sensitivity tests should be based upon high and low growth forecasts of background levels of economic and traffic growth as well as the development of interventions across the study area as a whole. This will present an indication of the robustness of the without-intervention forecast to the decision-maker that quantifies the scope of uncertainty in the study area and that will better inform the decision of whether or not to proceed with the recommendation of certain options. It may also be prudent, to carry out sensitivity tests to the intervention itself.

1.5.33. More detail concerning sensitivity testing is in The Treatment of Uncertainty in Forecasting (TAG Unit 3.15.5). Issues specific to other public transport modes on highways is in Forecasting and Sensitivity Tests for Major Public Transport Schemes (TAG Unit 3.15.3). Uncertainty regarding rail interventions is covered in Methodology for Forecasting Rail Demand (TAG Unit 3.15.4).

1.6. Reporting Requirements

1.6.1. It is essential that the methods used to forecast future traffic levels, and the assumptions made, are fully documented in a Forecasting Report.

1.6.2. The following list gives the basic requirements of the Forecasting Report:

• The changes to the base year network made to produce the core and senstivity test without-intervention scenarios;
• The development of future year planning scenarios and assumptions made to produce the core and sensitivity test without-intervention scenarios;
• The definition of with-intervention strategies and packages, ie the changes to the without-intervention network to produce with-intervention network;
• Traffic demand forecasts: methods and assumptions (including steps to make consistent with national forecasts);
• Sources and assumptions for updating of generalised costs (assumptions for value of time, vehicle operating costs; assumptions of public transport fares and related costs);
• Assessment of uncertainty and results of sensitivity testing.

2. Further Information

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

3. References

ATOC (June 2005), Passenger Demand Forecasting Handbook (version 4.1).
[The Passenger Demand Forecasting Handbook is restricted to members (and those carrying out analysis on behalf of members) of the Passenger Demand Forecasting Scheme. This scheme is administered by the Association of Train Operating Companies (ATOC)].

DfT (November 2007), English Regional Traffic Growth and Speed Forecasts, www.dft.gov.uk/pgr/economics/ntm/AF07_Annex_Baseline_summary.xls (MS Excel).

4. Document Provenance

This Unit and its associated family of Units in TAG 3.15 supersede the following information:

DMRB Section 12.1.12 (The forecasting section of the Traffic Appraisal Manual (TAM));

DMRB Section 12.2.1 (Traffic Appraisal in Urban Areas ): part 5, Traffic Forecasting.

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

Integrated Transport Economic Appraisal (ITEA) Division
Department for Transport
Zone 3/06 Great Minster House
76 Marsham Street
London, SW1P 4DR
E-mail: itea@dft.gsi.gov.uk
Tel: 020 7944 6176
Fax: 020 7944 2198