Rail track charges in Great Britain—the issue of charging for capacity

Rail track charges in Great Britain—the issue of charging for capacity

Chris Nash*, Simon Coulthard1, Bryan Matthews

University of Leeds, Institute for Transport Studies, 36-40 University Road, Leeds LS2 9JT, UK

Received 10 March 2003; revised 1 July 2003; accepted 1 December 2003

Available online 10 March 2004

Abstract

Separation of infrastructure from operations in rail transport raises the issue of what should be the structure and level of charges for

infrastructure use. This paper outlines the solution adopted in Britain, and how it has developed. It concludes that the principle defect in the

current system is the lack of any charge to reflect scarce capacity. A way of measuring the capacity requirements of different types of train

and of identifying their opportunity cost is put forward, but it is recognised that reflecting this in the charges would be complex.

q 2004 Elsevier Ltd. All rights reserved.

Keywords: Rail; Charges; Capacity

1. Introduction

When the government chose to privatise British Rail using

a structure that separated infrastructure from operations, the

result was a need to develop a method of charging for the use of

rail infrastructure. Given that this charging regime would be

one of the key incentives influencing the service level

decisions of train operators, it is obviously of crucial

importance to the privatised railway regime. Economic

principles suggest that train operators will only have the

correct incentives regarding speed and frequency of service

and type of equipment to use, if the change in the charge levied

as a result of a service level change reflects the marginal social

cost of that change. In the case of rail infrastructure that

marginal social cost will generally reflect wear and tear on the

system (leading to changes in maintenance and renewal costs),

any increased operating costs such as signalling, any external

costs of accidents and environmental effects, and costs of

increasing capacity utilisation, including increased delays to

other operators and scarcity costs (the inability of other

operators to get the slots they want).

The aim of this paper is to explain the arguments behind

the development of rail infrastructure charges in Britain to

date, to present a critique of the system as it now stands

and to offer some suggestions for further improvement. It is

argued that the key failure of the current system is its failure

to charge adequately for scarce capacity and a major part of

the paper will be devoted to proposals to overcome this

deficiency. Section 2 of this paper presents the history from

the original set of charges to the first periodic review.

Section 3 presents a critique of the outcome and discusses

ways to overcome its deficiencies, Section 4 presents

proposals regarding the development of appropriate

capacity charges and Section 5 illustrates these proposals

with a case study. Our conclusions are presented in Section 6.

2. The development of rail track charges in Britain

The original approach to rail access charges in Great

Britain was determined by the government prior to

privatisation and set out in the document Department of

Transport (1993). What this paper proposed was that freight

and open access operators should pay a negotiated charge, at

least covering their avoidable costs and making as large a

contribution as possible to fixed and common costs.

Franchised operators should pay a variable charge equal

to the cost implications of running additional trains, and a

fixed charge equal to their other avoidable costs plus a share

of fixed costs not covered by freight and open access

operators or other sources of revenue.

The aim of this structure was to reconcile the fact that the

majority of infrastructure costs were found to be common

0967-070X/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.tranpol.2003.12.003

Transport Policy 11 (2004) 315–327

www.elsevier.com/locate/tranpol

1 Simon Coulthard works in the rail industry; he was a part-time Masters

student at the Institute between 1999 and 2001.

* Corresponding author. Tel.: þ44-113-343-5325; fax: þ44-113-343-

5334.

E-mail address: [email protected] (C. Nash).

http://www.elsevier.com/locate/tranpol

between operators, and—at least in the short to medium

term—fixed (para 3.2) with a belief that efficiency of the

infrastructure provider would be promoted if all its costs had

to be covered from revenue from train operators. Para 1.3

illustrates this logic:

Where subsidy of the operation of railways is appropriate

on social grounds, it is more efficiently directed at

particular services and paid to the operator rather than to

the provider of the infrastructure.

However, this could not be done simply by raising

charges above marginal cost without major distortions to the

efficiency of use of the infrastructure (para 3.3).

If Railtrack were to charge all operators a proportion of

common and fixed costs through a standard tariff, it

would drive off the railways traffic which was in a

position to pay for its avoidable costs….

The recommended solution was therefore that:

The long term health of the railway industry will be best

secured if Railtrack pursues a policy of market pricing,

subject to the avoidance of unfair discrimination between

competing operators in the same market. All operators

should therefore pay the avoidable costs which can be

attributed directly to them, and should contribute to

common costs differentially, reflecting their ability to pay.

It became the duty of the Rail Regulator to review all

aspects of access agreements, including infrastructure

charges, and he first consulted on this in Office of the Rail

Regulator (1994a).

By this time it was clear that the proposed structure of

charges for franchised passenger operators was one in which

the short run implications of changes in service levels for

charges would be very small. On average, only 8% of the

total charge they paid would be variable in the short run, and

most of this was simply paying for electricity. Thirty seven

percent of the charge would be to cover the long term

incremental cost of meeting the operators’ need for

capacity, but the level of this would not vary, at least

during the 5 year review period. Around half of the total

charge would be an arbitrary allocation based on some

measure of output. Forty three percent of the total would be

an allocation of common costs (of which about half, arising

at below the zone level were allocated on the basis of

planned vehicle miles and half arising at national or zonal

levels were allocated on the basis of budgeted revenue). The

remaining 12% would be station and depot access charges,

which again were shared between operators using them

based on an arbitrary allocation formulae.

Where new access rights were required, the price for

these would be negotiated at a level in between avoidable

cost and the value of the path to the operator on the basis that

Railtrack would be entitled to a greater share of the revenue

the more it was bearing risk.

The Regulator did not propose any major changes to

structure, but did comment that different operators would

vary in the quality of the paths they required and in their

peakedness, and suggested the possibility of the variable

element of the charge varying according to these factors.

Thus, the fact that the structure of charges implemented

gave no incentives for the efficient use of limited peak

capacity was recognised very early on.

The Regulator’s conclusions on the structure of track

access charges were published in Office of the Rail

Regulator (1994b). Whilst the Regulator argued that it

would be desirable for a greater proportion of access

charges to be variable with use, he did not consider it

appropriate to change the structure of charges in the short

term. Instead, he introduced procedures for the renegotia-

tion of access rights and charges, in the hope that this would

give Railtrack an incentive to ‘buy back’ scarce capacity

where it could put it to better use.

The first periodic review of track access charges started

with the publication of a consultation document in

December 1997 (Office of the Rail Regulator, 1997). The

Regulator considered that charges should incentivise Rail-

track, train operators and funders to maximise the efficient

use and development of the network whilst avoiding undue

discrimination between operators, appropriately rewarding

Railtrack for changes in the level of output and meeting the

government’s overall transport objectives.

Arguably, existing charges were meeting none of these

objectives. Negotiations for freight and open access operators

were complex and time consuming, whilst negotiations on

variation of access rights for franchisees were simply not

working. Moreover, the ability of Railtrack to negotiate

charges according to the ability of a TOC to pay, led to extreme

secrecy about demand on the part of TOCs to the detriment of

service and investment planning. More crucially, the charging

structure for franchisees gave no incentive for economy in the

use of scarce capacity and no adequate mechanism for the

replacement of existing low value services by higher value

ones. Operators were not adequately charged even for wear

and tear, and not charged at all for congestion and opportunity

cost of slots. The problem was particularly acute since there

had been a rapid growth in both rail traffic and train service

levels. Partly this was simply recovering from the recession

but even after this recovery was complete over the period

1997/8 to 2002/3 there was a 14% increase in passenger

kilometres and a similar increase in passenger train kilometres

(Strategic Rail Authority, 2003b). There was also continued

growth in freight traffic (11% increase in freight tonne km

between these years). This led to much greater congestion and

requirements for investment in new capacity than had been

anticipated, and it was the policy of the new government that

this should continue.

During the review, Railtrack provided evidence of

substantially higher wear and tear costs than allowed for

C. Nash et al. / Transport Policy 11 (2004) 315–327316

in the existing charges, and also quantified congestion costs

in fine detail by track section and time period. (Gibson et al.,

2002). It should be noted that the direct delays caused by an

additional train, for instance due to locomotive failure, were

already charged for through the performance regime; what

was charged for here was the additional delays to

subsequent trains simply due to the train in question

taking up capacity and thus reducing the ability of the

system to recover from delays caused by other factors.

Consideration was given to improving the incentive of

Railtrack to expand the network by also incorporating the

capital costs of expansion into the variable element of the

access charge on the basis of a calculation of long run marginal

cost; however, it was found that this varied enormously with

the location, size and nature of the additional capacity required,

and no feasible way of including this in the tariff was found.

Instead, attention concentrated on quantifying the congestion

cost of adding additional trains to the network. Arguably this

was sensible, given the long time periods and indivisibilities

involved in many plans to upgrade capacity. The cost of

additional delays has been estimated by means of modelling

(Gibson et al., 2002). The approach taken by Railtrack was to

use historical data on delays and capacity utilisation to specify a

function that could replicate the observed delays. This involved

identifying appropriate measures of delay and of capacity

utilisation, identifying appropriate functional forms and then

testing the strength of the relationship between incremental

delay and capacity utilisation. The result was a proposed tariff

broken down into several thousand-track sections and by time

of day. However, the Regulator both simplified the structure

and halved the level of charges before incorporating this

elementofcosts into the tariff. It seems that he was concerned at

the degree to which levying the full congestion charge might

reduce demand (and it must be said that the proposed charge

was based on existing, rather than equilibrium, levels of

congestion. On the other hand, given the expected underlying

growth in demand, it may reasonably be expected that

congestion will get worse rather than better, and there is

certainly no good reason to suppose that it would halve). He

may also have considered that the value attached to delays in

the calculation was too high.

The recommendations of the Regulator at the end of the

process were (Office of the Rail Regulator, 2000, 2001):

† an increase in the variable part of the track charges to

reflect the full wear and tear cost and 50% of the

quantified congestion cost.

† a move to a published tariff for all operators, with

franchised operators continuing to pay on a two part

tariff, but freight and open access operators paying only

the variable element of the tariff. However, open access

continued to be heavily restricted, and the Regulator,

subsequently, suggested that where open access entry on

a hitherto protected flow was permitted, the entrant might

be required to compensate the franchisee for loss of

profits.

† an incentive payment to Railtrack based on increases in

traffic in order to encourage expansion of the network.

Because this was not funded through the variable part of

the track access charge, there was no corresponding

disincentive to train operators to expand, as there would

have been had train operators paid this directly.

In the event not all of the costs falling on Railtrack as a

result of these decisions were added to the fixed element of

franchisees’ charges; the Strategic Rail Authority agreed to

pay for the removal of the contribution to joint and common

costs from freight operators, and a part of the general increase

in Railtrack’s costs through direct payment to Railtrack. This

avoided a situation whereby franchisees’ fixed payments

would have increased, but under the terms of the franchise

agreements SRA would have had to compensate them for

these increases anyway. By entering into a direct financial

relationship with Railtrack, arguably SRA would have more

control on how the money was spent.

3. The existing structure—a critique

The modifications brought about by the first periodic

review brought the structure of rail track charges in Britain

much closer to the theoretical ideal. Wear and tear was more

appropriately charged for, with the charges varying in fine

detail according to the characteristics of the vehicle although

still not representing variation by track type. A capacity charge

was introduced based on the congestion cost caused by an

additional train, and although the Regulator simplified Rail-

track’s original proposal, it was intended that this should still

vary in fairly fine detail according to where and when the train

operated. In practice, it is understood that the capacity charge

only varies by train type, and not by time and place, because of

problems with the billing software. This is disappointing, but

we do not believe it reflects a fundamental problem with

moving to a more finely differentiated charging system.

We commented in the introduction that infrastructure

charges should also meet the external accident and

environmental costs of additional services. Given the low

accident risks, and the fact that railway companies are

responsible for their own insurance, it seems unlikely that

the external accident cost is very large. Environmental costs

have been quantified in a recent study (Sansom et al., 2001)

and are shown in Tables 1 and 2. Although, these are much

smaller for rail than road, it is clear that they are typically

significant relative to the marginal infrastructure usage costs

and should therefore be included.

One of the most difficult issues to deal with in rail

infrastructure charging is that of scarce capacity. Ideally

charges would give train operators appropriate incentives to

expand services only where the value of the service is at

least as high as the costs it causes, and where capacity is

scarce to ensure that it is used to provide the services of

greatest value. This issue has become of great importance

C. Nash et al. / Transport Policy 11 (2004) 315–327 317

given the growth of traffic and the high costs of expansion

and the consequent need to make the most effective use

possible of what capacity is available (Bowker, 2002).

We explained above how a capacity charge based on the

cost of congestion was introduced. But congestion is only the

appropriate capacity cost where the train in question

constitutes an additional train to what would otherwise have

been run; where the train in question runs instead of some other

train the appropriate capacity cost is the opportunity cost of

trains forced off the system by lack of capacity. Where

capacity constraints bite, use of a particular slot by one train

operator leads to inability of others to obtain their desired slots.

Where there is a choice between operating an additional train

and forcing another one off the system, obviously the course of

action with the lower social cost should be taken, so it will be

the lower of these two values that should be adopted.

Charging for scarce capacity would require estimation of

the opportunity cost of a slot. The most attractive solution to

this problem in theory is to ‘auction’ scarce slots. There are

many practical difficulties, however, including the compli-

cated ways in which slots can be put together to produce a

variety of types of service, and the fact that the value of

a particular slot for a particular use depends on how other

slots are being used (in terms of the operation of

complementary or competing trains). It is also the case

that the willingness to pay for the slot by the train operating

company will only reflect its social value to them if

appropriate subsidy regimes are in place to reflect the user

and non-user benefits of the service as discussed below. In

practice, it is therefore usually accepted that any degree of

price rationing of scarce slots will have to be on the basis of

administered prices rather than bid prices, although some

countries, including Britain, allow for a degree of

‘secondary trading’ in which slots change hands between

operators at enhanced prices (strictly, this must take place

through Railtrack, so it is not secondary trading in the sense

forbidden by EC Directive 2001/14). Nilsson (2002)

provides a more detailed consideration of auctioning.

A second possibility is to simply impose a price and see

what happens to demand, and then iterate until demand

equals capacity. The risk is, however, that serious distor-

tions may occur whilst the price is adjusting, and that

strategic game playing may occur to force the price down by

withholding demand, where competition is not strong.

A third approach, recommended by NERA (1998), is to

identify sections of infrastructure where capacity is con-

strained and to charge the long run average incremental cost of

expanding capacity. However, as explained above this is a

very difficult concept to measure (the cost of expanding

capacity varies enormously according to the exact proposal

considered, and it is not easy to relate this to the number of

paths created, since they depend on the precise number and

order of trains run). It may be argued, however, that more

appropriate incentives are given to infrastructure managers if

they are allowed to charge the costs of investment they actually

undertake, rather than for the scarcity resulting from a lack of

investment, at least, if they are commercially oriented. For

short run marginal cost pricing encourages them to restrict

capacity in order to keep price high; whereas a system

where a capacity charge reflected actual expenditure on

expanding capacity would overcome this problem. EC

Directive 2001/14 which governs rail infrastructure charges,

seeks to get round this by requiring infrastructure managers to

undertake studies to determine the cost of expanding capacity,

and to test whether this is justified on cost-benefit grounds,

where scarcity charges are levied.

Given the difficulties with all these approaches, it may be

thought that the best way of handling the issue is to permit

direct negotiation between operators and the infrastructure

manager over the price and allocation of slots, including

investment in new or upgraded capacity. However, British

experience of this approach is that it is complex, time

Table 2

Marginal cost and revenue analysis for passenger rail £/train km, low cost

estimates

Costs Category

Inter-city Regional London Passenger

sector

Marginal infrastructure

usage

1.116 0.149 0.406 0.424

Vehicle operating cost 11.79 5.04 6.68 7.07

Electricity 0.483 0.068 0.371 0.228

Congestion 0.15 0.09 0.28 0.18

Mohring effect 21.55 20.67 21.19 21.05

Air pollution 0.279 0.041 0.067 0.098

Noise 0.122 0.042 0.088 0.076

Climate change 0.067 0.031 0.037 0.040

VAT not paid 2.46 0.54 1.48 1.32

Total 14.92 5.33 8.22 8.38

Revenue 14.07 3.11 8.47 7.52

Difference cost-

revenue

0.85 2.22 -0.25 0.86

Note: Low cost estimates apply to environmental categories only.

Source: Sansom et al. (2001).

Table 1

Marginal cost and revenue analysis for rail freight £/train km, low cost

estimates

Costs Category

Bulk Other Freight sector

Marginal infrastructure usage 1.79 0.88 1.19

Vehicle operating cost 8.60 9.70 9.28

Air pollution 0.166 0.166 0.166

Noise 0.170 0.170 0.170

Climate change 0.131 0.131 0.131

Total 10.86 11.05 10.94

Revenue 13.01 13.61 13.41

Difference cost-revenue 22.15 22.56 22.47

Note: Low cost estimates apply to environmental categories only.

Source: Sansom et al. (2001).


consuming and will not necessarily lead to an optimal

outcome given the number of parties involved and the scope

for free-riding.

An alternative is for the track charging authority (or the

Strategic Rail Authority on its behalf) to attempt to calculate

directly the costs involved. For instance, if a train has to be

run at a different time from that desired, it is possible to

use studies of the value people place on departure time shifts

to estimate the value to its customers of the cost involved.

Similarly, the costs of slower speeds may be estimated from

passengers’ values of time.

Given the current degree of excess demand for slots, it is

likely that the failure to charge for scarce capacity, together

with the under charging for congestion and the exclusion of

certain other elements of marginal social cost, is leading to a

situation where slots are substantially under priced,

compounding the problem of capacity shortage by leading

to incentives to train operating companies to run too many

trains, to have too strong a preference for frequent short

trains rather than less frequent long, and to seek particular

timings that are wasteful of capacity. The importance of

these issues is illustrated in the case study below.

One counter argument should be considered first,

however. That is the argument that the infrastructure of

other modes, including road and air, are also not charged for

in a way that adequately reflects marginal social cost, and in

particular congestion and scarcity costs. Indeed, the study

referred to above (Sansom et al., 2001) found substantial

undercharging for the road mode on average, and too little

differentiation in the current charges (primarily fuel tax)

between locations and times where congestion is a problem

and those where it is not. It might therefore, be argued that

to charge rail operators for these costs when road operators

do not have to pay them is counter productive.

This is, however, a very simplistic view of the

appropriate approach to such ‘second best’ conditions.

The extent to which particular services divert passenger or

freight traffic from congested or environmentally damaging

roads or airports will differ with the type of service and how

heavily it is loaded. The appropriate way of dealing with

such second best considerations is, therefore, to pay train

operating companies grants to reflect the benefits elsewhere

of diverting traffic from other modes. The Strategic

Rail Authority does indeed already pay grants to freight

customers designed to attract traffic to rail in such

circumstances, and it and the Passenger Transport Execu-

tives do of course also provide financial aid for passenger

services. However, the grants do not at present vary with or

reflect the benefits of the attraction of additional passengers

to rail, and more could be done to improve incentives here.

4. Capacity charges—a proposed approach

The basis of this theory is that operators should be

charged for the capacity they use in accordance with the

social opportunity cost of that capacity. In order to

implement this approach, it is necessary first to measure

the amount of capacity used by each train run, and then to

estimate its opportunity cost. Both stages of the process are

very complex.

The big problem with measuring rail capacity is that the

capacity of a given stretch of railway line depends not just

on its physical characteristics (number of tracks, signalling

system, line speed) but also on the characteristics of the

trains using it, and in particular whether they are travelling

at different speeds, and the order in which they run.

Capacity (though not necessary benefit) is maximised, if all

trains have the same average speed; generally the more

diverse the speeds the fewer the number of trains that can be

accommodated. Moreover, all these factors will vary for

individual segments of the route and the services operating

will vary by time of day, requiring calculations to be done in

fairly fine detail.

The following two figures provide an illustration of this

point; clearly showing that the capacity of the same route

section can be greatly influenced by the differing charac-

teristics of train slots. Fig. 1 shows an allocation of capacity

where all train slots have uniform characteristics (average

speed, stopping pattern, etc.). In these circumstances, it is

possible to plan trains to the minimum headway (5 min apart

on this route section). The result shows the maximum

number of slots per hour over the route to be 12. This would

be the same whatever the uniform characteristics.

Fig. 1. Theoretical capacity allocation between Huddersfield and Stalybridge assuming uniform train slots.


Fig. 2 shows how the capacity available on the same

route section is greatly reduced when train slots of varying

characteristics are planned in a mixed pattern. The addition

of stopping passenger and freight train slots greatly reduces

capacity. In this scenario, the introduction of two stopping

passenger and one freight train has reduced the capacity of

the route by a half.

The result is that it is only possible to tell how much

capacity a particular service will consume in the light of the

other services operating on the route.

This suggests that a sensible way forward may be to

define, for each stretch of track, a prime user in terms of the

dominant type of train, and to consider one path for such a

train as a standard path. Other types of train will then be

considered in terms of the number of standard paths they

use. Whilst strictly, even this will depend on what types of

trains are running in the adjacent slots, it may be reasonable

to assume that—in the absence of knowledge to the

contrary–the adjacent slots will be occupied by trains

requiring standard paths.

As explained above, there are two reasons why a train

may require more than one standard path. The first is that it

operates at a different speed to other trains on the route in

question. The second is the precise times at which the trains

operate. The calculation of these two elements will be

illustrated in Section 5.

The second issue is the calculation of the opportunity

cost of a slot. This is also difficult. Indeed, it can only be

known precisely when the exact timetable has been

developed and the identity of the marginal train excluded

is known (Quinet, 2003). However, by that stage all

decisions as to the allocation of capacity for the current

period have been taken, and therefore, the capacity charge

cannot help to achieve an efficient allocation for that period.

What is needed, therefore, is a simpler approach that still

gives reasonable signals to train operating companies on

which they can base their longer term planning.

Where there is no shortage of capacity, and it is not

expected to be a problem within the planning period then of

course, the most efficient approach demands a capacity charge,

over and above any congestion charge, of zero. Thus, we are

only talking here of capacity constrained sections.

When considering services other than those of the prime

user of the route, it will often be sensible to measure

opportunity cost in terms of the value of an additional standard

path to the prime user. The prime user is usually a passenger

operator, and existing passenger demand forecasting models

(such as the industry standard model MOIRA or the more

detailed simulation model PRAISE) may be used to forecast

the impact on demand and revenue of the allocation of the

additional slot. In addition, it is necessary to forecast user

benefits not captured as revenue, and non-user benefits—in

particular the proportion of traffic diverted from road and the

saving in external cost that entails.

If the best alternative use of the slot is not an additional path

to the prime user this will of course understate the value of the

slot. If the identity of the best alternative use may be found then

the same procedure may be followed to find a more accurate

estimate of the opportunity cost of the slot. Note that if the

alternative use requires more than one standard slot, then the

value of the alternative train must be scaled down propor-

tionately to find the value of a standard slot. It should also be

noted that the opportunity cost might be not that the user in

question is forced off the system completely but that they get a

slower or otherwise less desirable slot.

For the prime user itself, the opportunity cost of the slot

is of course always going to be in terms of additional trains

of another type. Thus, it will always be necessary to

estimate the value of competing uses of the route in question

to implement this approach to capacity pricing.

5. Applying the proposed capacity charge in practice:

assessing the theory on the North Transpennine

Rail Route

To illustrate how the theory of scarcity charges might

work in practice analysis was conducted on the main North

Transpennine rail route between Leeds and Manchester

(Fig. 3) and was based on the Summer 2001 Timetable

Fig. 2. Theoretical capacity allocation between Huddersfield and Stalybridge assuming successive non-uniform train slots.


(with a later, updated examination of the Summer 2002

Timetable). This route has seen a dramatic rise in patronage

since the late 1980s; accompanied by modernisation of

rolling stock, improved frequencies and a change of main

station in Manchester. There have also been exogenous

factors that have influenced the growth on the route,

including: increased economic prosperity of Leeds and

Manchester, and growing congestion on the M62 (the main

Fig. 3. The North Transpennine Route.


competitor for transpennine travel). In addition, the rail

industry has been considering how to upgrade the route and the

services over it. The Strategic Rail Authority has been

sufficiently convinced by the growth of the passenger service

crossing the Pennines that it has commenced a process to

refranchise ‘Transpennine Express’ as a separate train operator

with a distinctive quality of service. It is primarily a two-track

route with occasional passing loops. We do not consider the

problem of assessing capacity charges for single track routes,

where the interaction between trains may be much more

complex (although, in Britain at least the pattern of services is

typically simpler and the capacity problems less severe).

For our assessment ,the route was divided into three

sections, each route section being assigned a category

according to the ‘prime user’ characteristics of the rail traffic

over the route:

Huddersfield–Stalybridge: high speed express route.

Piccadilly–Guide Bridge: slow speed suburban route.

Huddersfield–Leeds: high speed express route.

The Guide Bridge–Stalybridge section is ignored as being

a relatively low speed section with no capacity problems.

On the basis of these categories a benchmark can be set

for the maximum slots per hour over each route section; and

the minimum time consumed per slot for each route section.

(Fig. 1 provides an illustration of this process).

The benchmark for the three route sections is as follows:

Huddersfield–Stalybridge: 12 standard paths per hour

(mean headway 5 min).

Piccadilly–Guide Bridge: 15 standard paths per hour


Huddersfield–Leeds: 12 standard paths per hour


The issues examined can be summarised as follows:

† How allocation of capacity to non-standard trains con-

sumes more than 100% of the capacity of a standard slot.

† An examination of the impact on capacity allocation of

the aspiration to operate a regular, clockface pattern of

four transpennine expresses per hour.

† An analysis of the impact of increasing the number of

train slots allocated to freight services.

5.1. The impact of allocating capacity

to non-standard trains

To illustrate the impact of allocating scarce capacity to

non-standard trains an assessment was made of the peak

hour on each of the three route sections. The capacity

utilised on each route section was assessed using the

‘capacity utilisation’ measure devised by Gibson et al.

(2002). Essentially, this squeezes the trains together,

preserving their ordering, to find the minimum time period

in which that set of trains could be run if the precise times at

which they run did not matter.

5.2. Huddersfield–Stalybridge

The actual capacity allocated during the peak hour (Fig. 4)

on this route results in seven paths being allocated in a

period of 65 min, with an average headway of 9.3 min,

compared to the benchmark of 5 min.

Treating the actual train paths using the Gibson method

(i.e. squeezing them together to the minimum headway

apart) produces the results shown in Fig. 5

The seven slots can now be squeezed into 50 min, with

an average headway of 7.1 min. In other words, it would

be possible to provide a further two standard paths per

hour, if there were complete flexibility over the precise

times of these slots. However, the presence of the non-

standard slots is still reducing capacity from 12 trains per

hour to 8.45.

Another way of looking at this question is to consider the

opportunity cost of introducing a single non-standard path

into an otherwise uniform pattern of standard paths such as

is shown in Fig. 1. For the Huddersfield to Stalybridge route,

the introduction of a single stopping train among a uniform

pattern of express services creates an opportunity cost of

three standard paths. Inserting a single freight service in

the same manner creates an opportunity cost of four

standard paths.

Fig. 4. Actual path allocation in busiest hour between Huddersfield and Stalybridge in Summer 2001 Timetable (Source: Railtrack Working Timetable).


On the Manchester Piccadilly to Guide Bridge route the

issue of scarcity was not considered because the capacity

(even in the busiest hour) was clearly not fully utilised.

The opportunity cost of introducing a single non-

standard path into a uniform pattern of standard paths on

this route is two standard paths (stopping services). This

applies where the non-standard path is either an express or a

freight service.

Between Huddersfield and Leeds, the allotted slot

departure time pattern was responsible for an opportunity

cost of three prime user slots. The allocation of capacity to

non-standard slots was also responsible for an opportunity

cost of three prime user slots in the sample hour.

The opportunity cost of introducing a single non-

standard path (stopping service) into a uniform pattern of

standard paths on this route is three standard paths (express

services).

5.3. The impact of regularising train slot patterns

When we conducted our assessment of the Summer 2001

timetable there existed an aspiration to provide a regular,

even interval, clockface pattern of four transpennine express

services per hour between Leeds and Manchester. With the

introduction of the Summer 2002 timetable, four transpen-

nine express services were introduced (although not at even

intervals). The pattern of four express trains per hour

provides an interesting example of allocating capacity at the

expense of other services.

Fig. 6 shows how capacity has been allocated between

Huddersfield and Stalybridge in the busiest 2 h period of the

Summer 2002 timetable. Careful planning has ensured that

the introduction of the four express services has not

compromised the ability to operate an hourly stopping

service (and an additional peak hour local service from

Greenfield). The service pattern has also accommodated a

freight service, albeit with a long layover in the passing loop

at Marsden. The compromise is the inability to provide an

even interval express service. The planned interval between

express services is 11, 16 12; 13, 19 1

2; 10, 17 1

2; and 13 min.

The ideal would be a regular interval of 15 min. This is

explored further in Fig. 7.

Fig. 7 clearly shows the opportunity cost of planning a

15 min interval express service. All non-standard paths have

been lost, with the exception of the additional peak hour

service from Greenfield. Within this rigid train pattern, there

Fig. 5. Capacity utilisation method applied to the busiest hour between Huddersfield and Stalybridge.

Fig. 6. Actual allocation of track capacity between Huddersfield and Stalybridge in busiest two hour period (Source: Summer 2002 Working Timetable).


is no way of accommodating a transpennine freight service

without substantial investment in additional running lines, at

least between Huddersfield and Diggle. There is potential to

include a transpennine stopping service, but only by

requiring an additional 11 min layover at Marsden.

5.4. Providing greater access for freight services

The Government’s Ten Year Plan for transport includes a

target of increasing freight tonne-kilometres by rail by 80%.

Key to delivering this target will be the ability for major

routes such as the north transpennine to accommodate

increased freight train usage. To illustrate the potential

impact of increasing freight train access, we modelled a

scenario where the local stopping service was substituted by

freight services. The results are shown in Fig. 8.

Our analysis suggests that it is possible to substitute the

existing stopping passenger service with a freight train.

However, the freight service is required to wait in the

passing loop at Marsden for two successive express services

to pass. The four express services per hour have been

retained but the interval between them has to be further

compromised to intervals of 5 12; 24 1

2; 5 and 25 min.

The opportunity cost of expanding the allocation of capacity

to freight services is the loss of the stopping passenger

service and loss of anything close to a 15 min interval

express passenger service.

To minimise delay to freight services, it would be

preferable to operate these trains without the layover at

Marsden. The impact of inserting two direct freight services

into the regular pattern of standard (express) paths is to

consume all but three potential standard paths within each

hour. The opportunity cost of such an option is not only the

loss of the stopping service but also one of the remaining

express slots. The interval between the remaining express

services would now be 27, 5, and 28 min.

6. Estimating the opportunity cost of slots

The prime user of the Huddersfield–Stalybridge route

section is the inter-urban express passenger service. It may

be reasonable to assume that an additional service of any

other type would require a reduction of the number of slots

allocated to this service. Therefore, we need to know

what is the opportunity cost of a slot for this type of service.

Fig. 7. Allocation of capacity required to deliver transpennine express services at 15 min intervals, showing paths ‘lost’ from allocation in Figure 6.

Fig. 8. Theoretical re-allocation of track capacity to additional freight services between Huddersfield and Stalybridge.


This can be estimated as the sum of:

† the additional amount of traffic attracted to rail by the

presence of this train multiplied by the price it pays.

† the consumers surplus to rail users as a result of the

additional quality and capacity provided by the train.

† the savings of external costs to road users and the

public at large from the train attracting passengers

from road.

Less the train operating and infrastructure cost savings

from failing to run this train.

In this particular case, there is no problem with the

capacity of the service to handle the number of passengers

since, it would be possible to run fewer longer trains with

the same capacity (with some complexity, in that the

additional capacity may only be needed over the Manche-

ster–Leeds section, and it might be difficult to achieve this

without joining and splitting trains at these stations). Thus,

the relevant revenue and other benefits relate simply to the

effect of higher frequency services.

It has not been possible to obtain data to make specific

calculations for this section of route. However, some

suggestions may be derived from other studies. These are

Regional services and the figures in Table 1 suggest that on

average there is a shortfall of £2.22 per train kilometre in

comparing revenue with marginal social cost for regional

services. This is with the typical load for these services

being of the order of 40 passengers. The standard rail

demand forecasting methods in use in great Britain would

suggest that, for this type of service, a change in frequency

from 4 to 3 per hour would be the equivalent to an addition

to journey time of some 4 min; the typical current journey

time might be of the order of 60 min including the frequency

penalty. Given evidence that the journey time elasticity is of

the order of 0.9, the conclusion is that this reduction in

services might lose around 5% of the total traffic. Assuming

the train in question, being 1 train out of 4, carried 25% of

the total of the service, then the lost traffic would be as high

as 40 only if the total number of passengers per train were as

high as 200, certainly well above that to be found on these

services. A more reasonable assumption would be 100

passengers per train, giving a revenue loss of half that in

Table 2 and a shortfall relative to costs of £3.77.

As against this, we must consider the loss of consumers

surplus, which for 400 passengers per hour losing 4 min

might total some £272, or £4.53 per train kilometre

(assuming a value of time of 17p per minute for the sort

of mix of traffic found on these services and a route length of

60 km). Strictly, we should deduct from this the Mohring

effect since we are calculating the effect of changing

frequency directly, so the additional benefit is £3.86.

We also need to consider the external costs imposed by

any passengers diverting to road. There is evidence that for

inter urban train services about 60% of the passengers would

otherwise use car (Vicario, 1999). Thus, this service

improvement may be diverting some 12 passengers per

hour from car, or with an occupancy rate of 1.5 relieving the

roads of eight cars per hour. Assuming that the roads

concerned are 50% outer conurbation motorway and 50%

rural motorway, and that 50% of the traffic is in the peak,

Sansom et al. (2001) suggests a benefit of 18p per car km, or

£1.44 per train km. This leads to a net benefit of around £1.5

per train km. Given that to achieve these benefits a train has

to operate right through from Leeds to Manchester, a train

which deprives the service of such a slot even for part of the

route should be charged for the full route, i.e. 60 £ 1.5 or

around £90 per slot.

It should be noted that this is a day long average. During

the peak, it is likely that the benefits will be much greater

than this and in the off-peak lower. Stopping passenger

services may have higher benefits per train kilometre, at

least in the peak, if they relieve congested urban roads to a

greater extent. On the other hand, they often operate with

much lower loads of passengers per train, particularly in the

contra-peak direction, although, their need for a slot may

still be strong in order to get into position for a second peak

run. It should be noted that the value of a slot as a

positioning movement is something we have not considered

in this paper but it may be an important factor.

Given that this route is one where a major conflict

appears to exist between plans for passenger services and

the desire to expand freight services, it may be interesting to

do some similar exploratory estimates for the net benefit of

an additional freight service. Table 2 suggests that non-bulk

freight services are already earning revenue some £2.50 in

excess of marginal social cost. Assuming an average load of

900 tonnes that would otherwise use road and require 45

articulated goods vehicles, and on the same assumptions

about the roads from which this is drawn but assuming it is

in the off-peak, there is a benefit of 23.5p per vehicle km

from relieving the roads of this traffic. In other words, per

train km there is a benefit of £10.58 to add to the £2.5 above.

So the total benefit of the freight service is £13 per train km.

If it is travelling the same 60 km as the passenger service

this gives a value per slot of £780, but of course, it may be

going considerably further than that. There is little doubt in

this case, therefore, that space should be freed up to fully

meet demand for freight services even though they require

more than one standard slot per train run. On the other hand,

it is more likely that freight services may be retimed or

rerouted, as opposed to the traffic being totally lost, in which

case the relevant net benefit may be much lower than that

quoted above. Moreover, there is less certainty that a slot

allocated to freight traffic will actually be used given short

term fluctuations in traffic.

If it is the case that the marginal benefit of allocating

additional paths to freight is substantially greater than that

of additional paths to the prime user, then of course charging

the prime user the value of one of its own paths will not be

sufficient to encourage transferring marginal paths to the

alternative use. This situation would not arise if the timetable


had already been optimised, for then the marginal value of a

slot should be the same to any operator. But given the

indivisibility involved this is only a limited response even if

one has a lot of faith in the allocation body. An alternative

might be a return to the pre-privatisation approach whereby

a prime user bore all the costs of the infrastructure on a

particular route except those paths paid for by other

operators. For then, again provided that other operators

are subsidised to the extent of the net social benefits of their

services, the prime user would bear a marginal cost of using

a slot equal to the opportunity cost to the next best use.

It must be stressed again that the above calculations are

exploratory; since they do not utilise actual data for the route

in question, they must not be taken to imply particular policy

conclusions regarding that route. These simple examples

show that the data exists to estimate opportunity costs, but that

the information requirements are severe. The data in question

is not usually available to the infrastructure manager, though

in Britain it is available to the Strategic Rail Authority.

However, these information requirements are just the same as

those needed to determine an optimal allocation of slots; if

that exercise is undertaken then the information needed for

optimal pricing will already exist. Indeed in Britain, the

Strategic Rail Authority is increasingly using just such an

approach to determine how best to use capacity in its Capacity

Utilisation Studies (Strategic Rail Authority, 2003).

7. Conclusions

The system of charging for track access in Great Britain

has been substantially improved as a result of the periodic

review. Firstly, the usage element of the charge has been

increased to reflect more accurately the wear and tear an

additional train causes. Secondly, a capacity charge has

been introduced to allow for the delays created by running

an additional train, although the level of this included in the

variable element of the tariff is only half that measured by

Railtrack and variation by location and time of day has not

been implemented. But there are other reasons to suppose

that the variable element of the charge is still far too low.

Firstly, environmental costs are excluded. Although, these

are far lower for rail than for road, relative to the size of the

variable element of the charge these are still significant. But

more crucially, the current charges do not allow for the

opportunity cost of scarce capacity.

A method has been outlined for identifying the relative

charge to be levied for different types of trains when capacity

is scarce, to allow for the fact that different types of trains

have different capacity requirements, and for identifying the

opportunity cost of scarce capacity. The method is complex,

but since these are the same calculations as are needed to

identify the optimal use of capacity, and SRA is increasingly

undertaking this exercise then the data should exist to

estimate the relevant opportunity cost of capacity.

Although, the calculations shown are only illustrative

they reveal some important points. Firstly, the overall

benefits per train kilometre, at least in British circum-

stances, seem to depend heavily on the consumer’s surplus

and external benefits, so that an auctioning system, which

did not find a way of feeding these into the bids, would be far

from optimal. Moreover, the point that is often made that,

where the trains competing for a particular slot belong to the

same company the costs in question are internalised, is again

only valid if those companies are being subsidised

according to the net social benefits they produce.

Would it be worth introducing such a charge? It would be

necessary to identify and make calculations for all capacity

constrained sections of track and translate this into charges

varying by time of day, in much the same way as is the case

with the current congestion charge. Even so, the charges

would be crude, and could not be used for detailed

assessment of slot allocation; as commented above, they

arise out of an optimal allocation of slots rather than

determining it (Quinet, 2003). Their role would be rather to

demonstrate the opportunity cost of slots to companies in

their longer term planning, and in that way influence them

before they reach the stage of detailed timetable planning.

But of course for that purpose, other anticipated changes in

circumstances, including infrastructure enhancement or

changes to the level of benefit produced by marginal

prime user train would need to be taken into account.

Perhaps, given the extent to which Britain has now

moved towards a centrally planned system in which the

Strategic Rail Authority determines the timetable to be

operated using cost-benefit analysis, the importance of

charging appropriately for the use of scarce capacity is now

reduced. But we consider that, in any country where it is

desired to give train operating companies appropriate

incentives for the long term development of their services,

capacity charges along the lines outlined in this paper will

be needed.

Acknowledgements

The authors are grateful to participants at the eighth

international conference on competition and ownership in

passenger transport (Thredbo 8), to Prof Ralph Turvey of

the London Business School, Dr J Preston as editor of this

journal and two unknown referees for very helpful

comments on an earlier version. All errors and opinions

are of course due to the authors alone.

References

Department of Transport, 1993. Department of Transport Gaining access to

the Railway network. The Government’s proposals..

Bowker, R., 2002. Britain’s Railway—Time for a New Radicalism. (The

Sir Robert Reid Railway Lecture 2002). The Institute of Logistics and

Transport, London.

Gibson, S., Cooper, G., Ball, B., 2002. Capacity charges on the UK rail

network. Journal of Transport Economics and Policy 36 (2), 341–354.

NERA, 1998. An Examination of Rail Infrastructure Charges. NERA,

London.


Nilsson, J.-E., 2002. Towards a welfare enhancing process to manage railway

infrastructure access. Transportation Research A 36 (5), 419–436.

Office of the Rail Regulator, 1994a. Framework for the Approval of

Railtrack’s Track Access Charges for Franchised Passenger Services. A

Consultation Document.

Office of the Rail Regulator, 1994b. Office of the Rail Regulator Railtrack’s

track access charges for franchised passenger services. Developing the

structure of charges. A Policy Statement.

Office of the Rail Regulator, 1997. The periodic review of Railtrack’s

access charges: a proposed framework and key issues.

Office of the Rail Regulator, 2000. Periodic Review of Railtrack’s Access

Charges: Final Conclusions. ORR, London.

Office of the Rail Regulator, 2001. Office of the Rail Regulator Review of

Freight Charging Policy: Provisional Conclusions. ORR, London.

Quinet, E., 2003. Short term adjustments in rail activity—the limited role of

infrastructure charges. Transport Policy 1, 73–80.

Sansom, T., et al., 2001. Surface Transport Costs and Charges, Institute for

Transport Studies, University of Leeds.

Strategic Rail Authority, 2003a. Strategic Rail Authority Capacity

Utilisation Policy: Network Utilisation Strategy. SRA, London.

Strategic Rail Authority, 2003b. National Rail Trends. Fourth quarter

2002–2003. SRA, London.

Vicario, A.J.B., 1999. Diversion Factors and Cross Elasticities, Unpublished

MA Dissertation, Institute for Transport Studies, University of Leeds.


Rail track charges in Great Britain—the issue of charging for capacity

Documents

Transcript of Rail track charges in Great Britain—the issue of charging for capacity