When to invest in high speed rail
Chris Nash
Institute for Transport Studies, University of Leeds, United Kingdom
a r t i c l e i n f o
Article history:
Received 10 June 2014
Revised 15 February 2015
Accepted 16 February 2015
Available online 12 March 2015
Keywords:
High speed rail
Investment
Appraisal
a b s t r a c t
This paper starts by a general review of the costs and benefits of high speed rail, of how they are measured
in cost–benefit analysis and of the circumstances in which benefits may be expected to exceed
costs. Two approaches are taken to the latter; first, examining models in which values of key parameters
are varied to see in what circumstances benefits exceed costs, and secondly looking at the limited
evidence from ex post studies, mainly for France and Spain. We then turn to British experience of the
appraisal of HS2 – the proposed line linking London to Birmingham, Manchester and Leeds. It is concluded
that the main factors determining economic success for high speed rail projects are construction
costs, value of time saving per passenger and traffic volume and degree of congestion of existing transport
networks. The biggest uncertainty regarding the case for high speed rail surrounds the possibility of
wider economic benefits.
Ó 2015 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2. Costs and benefits of high speed rail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2. Costs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3. Traffic and revenue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4. Time savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5. Release of capacity on existing routes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.6. Diversion from other modes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.7. Induced traffic and wider economic benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3. In what circumstances will benefits exceed costs? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2. The modelling approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.3. Ex post evaluations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4. High speed 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2. Appraisal of HS2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1. Introduction
High speed rail (HSR) is usually regarded as comprising services
operating at 250 kmph or more, and these speeds invariably
require construction of new purpose built lines, although high
speed trains may run at up to 200–225 kmph on upgraded existing
lines. According to UIC, by 2013, a total of 21,472 km of new high
speed lines had been built worldwide. China had the largest network
at 9867 km, whilst Japan, France and Spain all had over
2000 km. There are plans for a further major expansion, with the
http://dx.doi.org/10.1016/j.jrtpm.2015.02.001
2210-9706/Ó 2015 Elsevier Ltd. All rights reserved.
Journal of Rail Transport Planning & Management 5 (2015) 12–22
Contents lists available at ScienceDirect
Journal of Rail Transport Planning & Management
journal homepage: www.elsevier.com/locate/jrtpm
European Commission calling for a trebling of the kilometres of
high speed line in Europe by 2030. Yet high speed rail is an enormous
investment, and it is necessary to consider very carefully in
what circumstances such an outlay is justified.
The first such line, the new Tokaido line in Japan, was clearly
built with the twin aims of giving large time savings (and thus
competing effectively with air transport) and relieving capacity
constraints on the existing railway line. These were also clearly
the motives behind the construction of the first TGV line from
Paris to Lyons in France. But since then, wider motives have
appeared, including reducing carbon emissions by diverting traffic
from air and road, and promoting economic regeneration and
growth. The first part of this paper will consider at a general level
the costs and benefits of high speed rail, and evidence to date on
what determines their magnitude. We then examine evidence on
the circumstances in which benefits will exceed cost, drawing both
on theoretical modelling exercises and on ex post studies of actual
schemes.
We will then consider specifically the current debate on high
speed rail in Britain. The large volumes of existing rail traffic, and
the ability to serve many of the main cities of Britain with a single
line splitting into two branches, appear to make Britain an ideal
country for high speed rail. However, the proposed line from
London to the North, HS2, has proved very controversial. The final
section of this paper draws conclusions.
2. Costs and benefits of high speed rail
2.1. Introduction
The principal costs and benefit of HSR are listed in Table 1.
Section 2.2 discusses costs, Section 2.3 patronage and revenue,
Section 2.4 time savings, Section 2.5 rail capacity benefits, Section 2.6
diversions from other modes and Section 2.7 wider economic
benefits.
2.2. Costs
Table 2 summarises construction costs of high speed rail in
Europe and Asia. It will be seen that there is a considerable range
both within and between countries. Obviously there are many
factors influencing this including differing labour and land costs,
whether the line is designed for both freight and passenger trains
or passenger only and whether the line uses slabtrack or traditional
ballasted track. But particularly significant factors are the
degree to which new stations have to be constructed and the
length of tunnelling needed (either because of the terrain or to
alleviate environmental problems (SDG, 2004). For instance of
the estimated construction cost of HS2 Phase 1 in Britain, over
40% comprises station and tunnelling costs (HS2, 2012). This is
because a substantial amount of tunnelling is planned, particularly
in order to get into central London, and either new or substantially
extended stations are required. By contrast, in many European
cities (such as Paris) high speed trains have generally been accommodated
in existing stations without major extensions and
accessed these on the surface. One reason for this is that suburban
trains have been redirected into underground routes across city
centres releasing surface capacity for high speed rail.
Given these high costs, it is obviously sensible to ask whether
the costs may be reduced by using upgraded conventional tracks
for all or part of the network. Much will depend on what already
exists and whether it has spare capacity. For instance, in Japan
the existing lines were metre gauge with many curves and gradients
and not easily upgraded for high speed operation. In Spain,
existing lines were broad gauge and difficult to upgrade for high
speeds, but the decision to build a new network of standard gauge
lines has limited through running to the conventional network to
trains capable of changing gauge. Even in parts of Europe, where
the existing lines were much more suitable, upgraded lines are
not usually used for more than 200 kmph (although 225 kmph
might be feasible with new cab signalling). A further consideration
is capacity; if lines upgraded for higher speeds have to accommodate
existing trains at lower speeds (e.g. freight or regional passenger
trains), the increased spread of speeds will actually lower
capacity.
What is more common is to build new lines where demand is
strongest and capacity most scarce but to use existing lines elsewhere.
The French TGV network has been developed along these
lines, with new tracks being built for the trunk haul into Paris
but trains proceeding to a host of destinations on the old network
(some of these destinations are now on extensions of the high
speed network). The German ICE network uses upgraded conventional
lines for much of its distance with new lines largely confined
to overcoming bottlenecks.
In terms of operating cost, it appears that high speed trains are
no more expensive than conventional trains when the capital cost
of the vehicles is taken into account. Whilst energy consumption
and maintenance costs are higher than for conventional trains,
high speed means staff and rolling stock can achieve much higher
utilisation rates than conventional rail, offsetting the increased
costs (Civity, 2013).
Of the external costs of high speed rail projects, noise, global
warming and loss of amenity through land take and visual intrusion
are the major issues. Noise costs and loss of amenity can be
minimised at the expense of additional capital cost, ultimately by
tunnelling.
Of these costs greenhouse gases has proved particularly contentious.
Other things being equal, energy consumption rises
rapidly with speed, particularly at speeds above 300 kmph, so high
speed trains may be more energy intensive than conventional
trains. Moreover the higher speeds will induce additional traffic
that has not simply diverted from other modes. Offsetting this is
the fact that high speed trains typically run at much higher load
Table 1
HSR costs and benefits.
Costs Benefits
Capital costs Revenue
Operating costs Time savings (beyond those recovered in
higher prices)
External costs Release of capacity on existing rail routes
Loss of tax revenue (from traffic
diverted from road to rail)
Diversion from other modes – reduced
congestion, accidents and environmental
costs
Opportunity cost of public
sector funds
Induced traffic
Wider economic benefits
Table 2
Construction costs of high speed rail (2005 prices).
Euros m per route km
China 5.7–18.8
Belgium 16.1
France 4.7–18.8
Germany 15–28.8
Italy 25.5
Japan 20–30.9
Korea 34.2
Spain 7.8–20
Taiwan 39.5
Source: Derived from Campos et al. (2009) Graph 1.3, except China, which is from
Wu et al. (2014).
C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22 13
factors than conventional trains as a result of long nonstop runs
between major cities and the use of yield management systems;
load factors as high as 70% have been claimed for the Eurostar
trains between London and Paris and Brussels, and for the French
TGV network, as opposed to 40% for conventional trains. The result
of these differences in load factor is that the carbon cost per passenger
km for a Eurostar train, for instance, is very similar to that
for a 200 kmph pendolino (CILT, 2011).
As well as the carbon produced by train operation, there is
obviously carbon used in construction of the line, although in
Europe, to the extent that this will mainly be undertaken by industries
which are part of the European emissions trading scheme, it
may be argued that the cost of offsetting any increase in carbon
from this source will already be included in the capital costs of
the project. The same argument may be applied to electricity for
traction (and jet fuel when the emissions trading scheme is fully
implemented for aviation).
Putting this argument aside, the greenhouse gas emissions for
HSR operation depend very much on the source of primary energy
used to generate the electricity, and as electricity generation is
decarbonised this will go down. It is the source of the marginal
electricity generated as a result of the increased demand for
electricity generated by HSR that is of interest and this may not
be the same as the average source at that point in time.
The safety record of rail overall is very good, but for high speed
trains it is particularly good. New high speed lines are invariably
built with cab signalling and completely segregated from road
and pedestrian traffic (i.e. without level crossings). The only fatal
accident to date on a newly built high speed line was on the
250 kmph Ningbo–Taizhou–Wenzhou line in China, when two
trains collided at low speed due to human error following a signal
failure. There have been several serious accidents involving high
speed rolling stock running on conventional lines, however,
notably in Germany and most recently Spain.
The majority of high speed lines in the world have not been
commercially viable, and have needed government grants towards
construction costs (and sometimes operating costs as well). The
major exceptions seem to have been the first French and
Japanese lines (see Crozet, 2014 and Kurosaki, 2014). Added to this
is the fact that they actually cost the government tax revenue if
they divert trips from heavily taxed road transport to more lightly
or untaxed rail. (Of course, one reason for the heavier taxation of
road transport is its greater external costs, but these are best taken
into account in the cost–benefit analysis directly, rather than by
assuming that the level of taxation correctly reflects them). The
treatment of the cost of reduced tax to the government as one of
the costs of rail schemes has been a contentious issue in the UK.
The logic is that if the benefits of reduced congestion and pollution
from road traffic are included as a benefit in the appraisal in full,
then the loss of revenue that road transport would have paid in
the form of fuel and other taxes must be seen as a cost.
Given that taxation typically involves the deadweight loss of
distortions, (and indeed that for macro economic or political reasons
there may be overall budget constraints on the public sector)
the call on public funds may have an opportunity cost in excess of
the simple cash flow. For many years this opportunity cost was
allowed for in the discount rate applied to public sector projects,
which was based on the rate of return such funds could achieve
in the private sector. More recently, discount rates for public sector
projects have typically been much lower (for instance 3.5% in
Britain, reducing to 3% after 30 years). This strongly favours
projects with high capital costs and long lives, by increasing the
present value of the benefits more than the costs. However, it
raises the issue of how to allow for this opportunity cost. Some
countries (e.g. Sweden) shadow price public sector funds by
multiplying them by a shadow price greater than one (often 1.3).
Others such as Britain require a ratio of net benefits to public
sector funding greatly in excess of one – in Britain typically at least
1.5 – to allow for the opportunity cost of funds.
A further possible reason for a higher discount rate is to allow
for risk, but this is a crude way of doing so. Generally the favoured
approach now is by means of a quantified risk assessment, which
considers the probability distributions for all the elements
contributing to cost and uses these to build up a probability
distribution of the cost of the project in total. The decision taker
may then choose if they wish to be cautious and take a cost figure
higher than the 50th percentile of the distribution in the appraisal.
2.3. Traffic and revenue
The construction costs of high speed rail are largely fixed
regardless of traffic. High speed rail almost invariably requires a
double track main line with cab signalling. If all trains are identical
in performance and leave the main line at high speed turnouts
before slowing down to stop at any intermediate stations, then
in principle operation at 3 min headways is feasible, offering
20 trains per hour. Some margin to recover from delays is necessary,
but already France runs 13 trains per hour in the peak
between Paris and Lyons and Japan 15 between Tokyo and
Osaka. Britain plans a peak service of 18 trains per hour on HS2.
France is already operating trains with pairs of double deck units
offering around 1000 seats. Only the costs of rolling stock, stations
and depots vary significantly with traffic volumes. Thus high speed
rail systems have very high fixed costs which can only be justified
by high traffic volumes.
The world’s first high speed line, from Tokyo to Osaka, carried
39 m passengers in its first full year of operation (1965) and this
had grown to 149 m by 2008 (Albalate and Bel, 2012) with the help
of extensions to the line. Paris–Lyon opened with 19.2 m
passengers in 1985, whilst the following Atlantic, North,
Connection, Rhone-Alpes and Mediterranean lines all opened with
similar numbers (Paix, 2010). However not all lines have attracted
traffic in these sorts of volumes. At the other extreme, the
Madrid–Seville line opened with as few as 2.5 m passengers, and
Madrid–Barcelona with only 5.0 m (Sanchez-Borras, 2010).
Volumes of the necessary size may be obtained by linking
individual very large cities (e.g. Paris and London) or by linking a
chain of large cities so that flows between different cities are
aggregated together and trains remain busy throughout the route
(the so called ‘string of pearls’). Japan is clearly a case of the latter,
with 127 m people living at very high population densities mainly
in large cities along the coastal strip. France also is able to benefit
from this sort of geography to a degree, partly by using the ability
of TGVs to run at reduced speed on conventional lines to serve
additional cities. For instance, trains on the original French TGV
line from Paris to Lyons, went on to serve cities such as Avignon,
Marseilles and Nice (since then the high speed line has been
extended to serve these places directly). By contrast, Spanish cities
are smaller, and arranged around Madrid in a ‘hub and spoke’
pattern, requiring a different line to link each city to Madrid.
Clearly the volume attracted is not simply a matter of the population
but also its propensity to travel and the competitiveness of
rail with other modes. An upgraded conventional rail system may
achieve a commercial speed of the order of 160 kmph, whilst for
high speed rail designed for 300 kmph, 240 kmph may be feasible.
The journey time for each in terms of hours for certain distances is
shown in Table 3. However, competitiveness in terms of door to
door journey time also depends on access to the station and
frequency of service. Car provides door to door service with no
waiting time or schedule delay. For a door to door journey largely
on motorway, a 100 kmph average may be feasible. If rail involves
an additional 2 h in terms of access, egress and waiting time then
14 C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22
even high speed rail will only be faster than car for journeys of well
over 300 km. If part or all of the road journey is on congested or
low speed roads, then rail may be competitive over much shorter
distances, and even with upgraded conventional lines. Indeed for
shorter journeys, a frequent service calling at easily accessible
stops may be preferable to a high speed service calling only at
the very busiest stations and made up of high capacity less frequent
trains, with other passengers having to change into the
trains at these stops.
In comparison with air, it has been argued that – because for
most passengers it involves less access, egress and waiting time
– rail can compete with air provided the station to station rail journey
is not more than 3 h. Table 4 shows that indeed rail generally
has more traffic than air in these circumstances. (The exception is
Madrid–Barcelona, where in 2009 airlines had retained slightly
more than 50% of the market partly by offering a shuttle service
on which passengers may take the next available flight without
reservations. On the other hand, there are cases such as Paris–
Brussels, where air has completely withdrawn from the market
and books its passengers into a special seating area in the trains.)
Thus whilst upgraded conventional rail may be able to compete
at up to 450 km high speed rail pushes this up to at least 700.
However Table 4 reveals a diversity of rail share for the same
rail journey time. A major factor here is the geography of the catchment
areas. In a dense city with high quality public transport
focussed on the city centre and frequent inter city services, access,
egress and waiting time for rail may typically add less than 2 h to
the door to door journey time, whilst for a more remote airport,
including time spent at the airport, for air it may be at least three.
In a low density city, with weaker public transport, the advantage
of rail in access and egress may be much lower.
But competitiveness does not solely depend on journey times.
Rail may also have the edge over other modes on comfort and on
board facilities (for instance, the ease of use of devices such as
laptops, tablets and mobile phones). Whether fares will be
competitive depends on the pricing policy, but with increasing
use of yield management high speed rail fares start to look much
more like airline ones. Whether rail is competitive with car on
price obviously depends on factors such as the price of fuel, the
presence and level of road tolls and how many people are
travelling together.
2.4. Time savings
Table 3 in the previous section shows that substantial time savings
may be made by passengers previously using even upgraded
conventional trains. For car or air, the time saving depends very
much on the length of journey; if the journey is close to the breakeven
point between the modes then the time saving may be much
smaller. However, it is not always appreciated that many aspects of
comfort and convenience may also be included in the value of
(generalised) time used in economic appraisal. The raw values used
in current British appraisals are shown in Table 5. For leisure and
commuting journeys, these values are based on extensive revealed
and stated preference evidence on what people are willing to pay
to save time. However higher values are used for waiting time,
for time standing in a crowded train, for time spent walking to
access trains and for late arrivals (the evidence is that people are
willing to pay something like twice as much to save time walking
and waiting and three times as much to avoid being an hour late as
they are to save an hour in scheduled journey time) (Wardman,
2004). Around half of the benefits to rail users from HS2 is
estimated to come from reduced crowding, improved reliability,
reduced walking and waiting times and improved access and
interchange, as opposed to simple reductions in in-vehicle time
(DfT, 2013).
Of course in lower income countries such as China it must be
expected that the willingness to pay to save time, will also be
lower, and it will be more difficult to justify high speed rail. Wu
et al. (2014) conclude that in China it is only the busiest high speed
lines on the prosperous East Coast that are justified; many of the
other lines would be better built as conventional mixed traffic
railways.
Where passengers are travelling on business, it is assumed that
the benefit of faster journeys goes to the employer, not the
employee. The approach taken to this in Britain, as in the appraisal
systems of most countries, is to assume that this is to be valued at
the wage rate of the staff concerned, plus an allowance for the
overhead cost of employing labour. In a competitive market, this
will equal the value of the marginal product of labour, and thus
represents the value of the additional output produced when
labour is released from its current occupation. It also represents
the cost saving to the existing employer, and thus is a key input
into models to estimate land-use transport interactions or wider
economic effects of transport investments.
Valuing rail business travel time savings in this way has been
widely questioned in recent years. Firstly, it has been noted that
business travellers can and do work on trains, and that improved
information technology has made this easier and more productive.
According to a recent survey a third of rail business passengers in
Britain state that this is how they spend much of their travel time
(Lyons et al., 2007). However, Batley et al. (2013) point out that it is
not how people spend their time on average that matters, but how
this would be affected by a marginal change in travel time, and
Table 3
Comparative journey times for upgraded conventional and high speed rail.
Distance Hours Hours
(km) at 160 kmph at 240 kmph
200 1.25 0.83
300 1.875 1.25
400 2.5 1.67
500 3.125 2.083
600 3.75 2.5
700 4.375 2.917
Table 4
Rail share of rail/air market and rail station to station journey times.
Corridor Year Travel time Rail share (%)
Paris–Brussels 2006 1 h 25 min 100
Paris–Lyons 1985 2 h 15 min 91
Madrid–Seville 2003 2 h 20 min 83
Brussels–London 2005 2 h 20 min 60
Tokyo–Osaka 2005 2 h 30 min 81
Madrid–Barcelona 2009 2 h 38 min 47
Paris–London 2005 2 h 40 min 66
Tokyo–Okayama 2005 3 h 16 min 57
Paris–Geneva 2003 3 h 30 min 35
Tokyo–Hiroshima 2005 3 h 51 min 47
Paris–Amsterdam 2004 4 h 10 min 45
Paris–Marseilles 2000 4 h 20 min 45
London–Edinburgh 1999 4 h 25 min 29
London–Edinburgh 2004 4 h 30 min 18
Tokyo–Fukuoka 2005 4 h 59 min 9
Source: Compiled by the author from Campos et al. (2009), Sanchez-Borras (2010)
and SDG (2006).
Table 5
Values of time used in British rail appraisals (£ per hour, 2010 prices and values).
Business 31.96
Commuting 6.81
Leisure 6.04
Source: DfT (2013).
C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22 15
whether the time is used as productively as time spent in the
office. Secondly, business journeys often start and finish in unsocial
hours and it is not clear that all time saved on such journeys will be
used productively. Hensher (1977) developed a method for adjusting
business values of time to allow for these factors, but one
which is very demanding in terms of the information needed to
apply it. The Hensher approach would generally reduce the value
of business travel time compared to basing it on the wage rate plus
overheads; depending on the precise assumptions, Batley et al.
found cases where it might be as low as 32% of the existing
valuation.
On the other hand, empirical investigations using evidence from
both revealed and stated preference studies tends to suggest a
value at least as high as currently assumed, with values of time
being much higher for first class travel than economy (Wardman,
2004). Possible reasons for this are that employers perceive benefits
from staff not being obliged to work such long days and thus
being less tired, from not having to compensate staff for unsocial
hours as part of their remuneration package and from being able
to fit more meetings into a day, thus saving further travel or the
cost of overnight stays (Marks et al., 1986).
2.5. Release of capacity on existing routes
Building a high speed line which will divert a substantial volume
of traffic from an existing route not only creates a huge capacity
for high speed traffic itself; it may also permit growth of other
types of traffic on the existing line. Removing all the fastest services
from a route will reduce the spread of speeds and thus release
more capacity than simply the number of paths formerly taken by
the diverted trains. In Europe, a major incentive for the construction
of high speed lines has been to release capacity for more
freight trains on paths that involve less time in loops waiting for
faster trains to pass. If the new line is built into the centre of the
cities it serves, then it will also release capacity sufficient for an
expansion of commuter and other regional passenger services.
Of course, building a high speed line is not the only way of
achieving this. Building a new mixed traffic or freight dedicated
line (as is underway in India) will also create capacity for other services.
More limited upgrading (for instance grade separated junctions,
longer passing loops, etc.) will create some extra capacity,
whilst a decision may be taken simply to price the peaks in traffic
off rail. But if a new line is to be built on routes with a lot of passenger
traffic, in many cases in wealthier countries it has been
found that it is better to build a high speed passenger line as the
increase in cost involved (20% in the case of HS2 in Britain) is less
than the benefit it brings. As noted above, this does not necessary
apply to lower income countries such as China.
2.6. Diversion from other modes
The benefits of diversion from other modes take the form of
reductions in externalities – congestion, accidents and emissions.
Against these must be set any excess of mode specific taxes and
charges over and above marginal cost of infrastructure provision
and maintenance. What is needed then is firstly an estimate of
how many high speed rail travellers have diverted from each
mode. Once the mode from which passengers have switched is
known, information is then required on the marginal social cost
of the modes in question. Based on the evidence cited in Table 6,
reduced noise and pollution costs from air transport might provide
a benefit from switching to HSR of the order of 9 m euros at year
2000 prices for 1 m passengers diverted from air in 200 seater
aircraft on an 800 km route. Thus it would only be if there was a
very large diversion from air, probably on a route where air had
previously totally dominated the market, that this would make a
significant contribution to the benefits of HSR.
For car, the evidence of GRACE (2005) is that it is only when
there is substantial congestion that marginal social cost exceeds
charges (Table 7) for using cars in European conditions. Whilst
we might now believe in a rather higher cost for global warming
than applied in 2005, this would not change the conclusion. But
that raises the issue that if sufficient traffic is diverted to significantly
reduce road congestion, new demand will be generated to
occupy some of the space. Thus accurate results can only be
achieved by use of a full multi modal model. The additional road
traffic would then be the benefit rather than reductions in
congestion.
Benefits should also include the net benefits of diverting traffic
into the capacity released on existing rail lines. This may include
commuter journeys into large cities and heavy freight. Since these
are both types of traffic where the evidence is that external cost
typically exceeds charges, the benefits may be more substantial
(Greengauge, 2012).
2.7. Induced traffic and wider economic benefits
For induced traffic, the standard argument is that, since the person
was unwilling to travel at the previous generalised cost and is
willing at the new, the benefit must lie somewhere between that
derived by an existing passenger and zero; assuming a linear
demand curve the benefit will be half that accruing to an existing
passenger. However, if new trips are generated for leisure, commuting
or business, they may imply a shifting of economic activity.
Whether they also may imply increased economic activity is the
source of much debate. The long held position amongst most
cost–benefit analysts has been that, following Mohring and
Table 6
Air transport externalities.
Air pollution Climate change
Flight distance (km) Direct emissions Direct emissions Indirect emissions
(a) Externalities air (eurocents 2000 per passenger km)
<500 km 0.21 0.62 0.71
500–1000 0.12 0.46 0.53
1000–1500 0.08 0.35 0.40
1500–2000 0.06 0.33 0.38
>2000 0.03 0.35 0.40
40 seater 100 seater 200 seater 400 seater
(b) Noise costs per landing or take off at Schiphol (euros 2000)
Fleet average 180 300 600 1200
State-of-art 90 150 300 600
Note: Indirect emissions are the climate change and air pollution cost of the production and transport of fuel for air transport. Obviously there may be offsetting costs for rail
which need to be included in the cost of the high speed rail project.
Source: Infras et al. (2008).
16 C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22
Williamson (1969), in a perfectly competitive economy, there will
be no benefits of transport investment over and above the direct
user benefits that are measured in a standard appraisal. Whilst a
transport investment may change relative prices and lead to
expansion and contraction of other industries according to the
degree to which transport is an input to them, as long as price
equals marginal cost in those sectors, there will be no net benefits
of these changes, whilst in the absence of involuntary unemployment,
there will be no net benefits of job creation or removal.
Transport investments may change property prices, but this will
simply be a capitalisation of the benefits received by the users; it
will transfer benefits from users to property owners but have no
impact on the overall net benefit of the scheme.
More recently the shortcomings of this position have been
exposed to greater scrutiny. Firstly, it is clear that perfect competition
is not the norm. If transport improvements lower the cost of
production and thus encourage output to expand, to the extent
that price exceeds marginal cost for the good in question, there
is an additional benefit to take into account. Secondly, if improved
transport induces people to enter the labour market or to work
longer hours, whilst the benefit to themselves may be reflected
in their willingness to pay in the transport market, there is an additional
benefit to government in terms of the extra tax they pay.
Thirdly is the issue of agglomeration externalities. It appears that
there is a direct link between accessibility and labour productivity,
perhaps because in a larger labour market workers are better fitted
to the jobs they do, innovations spread more quickly and there are
economies of scale leading to better supply of business services.
These issues were introduced into the debate on rail investment
regarding the Crossrail project in London (Venables, 2007) and the
empirical work to quantify them was led by Graham (2007).
However, work by Graham and Melo (2010) for the HS2 project
in Britain concluded that whilst these impacts might be important
for conurbations, they were unlikely to be significant for intercity
passenger rail transport, because they are a small proportion of
total work trips and rail has a small market share. However, the
fact that intercity rail is heavily used by managerial and professional
people may mean that the market share understates its
importance, as these are exactly the sort of people for whom
agglomeration effects are most likely to be significant.
More recently, further work undertaken by Rosewell and
Venables (2013) for HS2 concluded that there might be a significant
benefit to the economy from increased specialisation as a
result of better connectivity. Moreover, to the extent that high
speed rail tends to centralise economic activity in large cities, there
may also be increased agglomeration benefits from land use
changes.
The only attempt to quantify these benefits for HS2 was undertaken
by KPMG (2013). The first step of their methodology is to
calculate the impact of HS2 on labour and business connectivity
by location. They do this by looking at journey times weighted
by the distribution of existing journey lengths for the purpose in
question (the so-called distance decay function). As seen in
Table 8, most areas gain, even when not directly served by HS2
(long distance journeys may still use HS2 for part of their route,
whilst other places gain from improved services using the capacity
released on the existing network), although it is the Midlands and
South Yorkshire that enjoy the greatest gains (they gain for journeys
to London but also to other cities in the Midlands and North).
They then regress labour productivity on rail connectivity and
similar measures of connectivity by road. The difficulty faced is
that these indicators of connectivity are all highly correlated (and
may be correlated with other benefits of a city centre location).
The result is that only one measure of connectivity can be included
in a single regression. They therefore run separate regressions of
labour productivity on rail connectivity and on road connectivity.
They scale down the parameter value on rail connectivity by
assuming that rail connectivity is responsible for a proportion of
the benefits equal to the ratio of the parameter values on rail connectivity
and road connectivity in the separate regressions. This
leads to an estimate that HS2 could add £15b p.a. to GDP. Whilst
this is not entirely additional to the current appraisal benefits, it
must represent a substantial uplift. However, as Overman (2013)
points out, there is no theoretical justification for the assumption
by which the rail share of the effect is estimated, and a sensitivity
test using mode share data to perform this allocation gives a much
lower value.
3. In what circumstances will benefits exceed costs?
3.1. Introduction
There are essentially two approaches to answering this question.
The first is to construct a model and explore the values of
Table 7
Long distance car trip externalities and charges (1998 euros per vehicle km).
Wear Congestion Environment Accidents Total cost Charges
Route 1
Peak 0.016 0.147 0.013 0.015 0.191 0.132
Off peak 0.016 0.002 0.017 0.015 0.050 0.132
Route 2
Peak 0.032 0.194 0.010 0.008 0.244 0.156
Off peak 0.032 0.003 0.014 0.008 0.156 0.056
Route 3
Peak 0.019 0.123 0.011 0.008 0.161 0.114
Off peak 0.010 0.002 0.015 0.008 0.044 0.114
Route 4
Peak 0.020 0.122 0.015 0.006 0.163 0.078
Off peak 0.025 0.002 0.020 0.006 0.048 0.078
Note: Route 1 is Milan-Chiasso, Route 2 is Chiasso-Basel, Route 3 is Basel-Duisburg and Route 4 is Duisburg-Rotterdam.
Source: GRACE (2005).
Table 8
Average change in connectivity by region in 2037 after investment in HS2.
City regions Change in labour
connectivity by rail (%)
Change in business
connectivity by rail (%)
Derby–Nottingham 14.7 23.2
Greater Manchester 1.4 18.8
Greater London 6.9 8.8
South Yorkshire 31.8 22.5
West Midlands 15.7 21.1
West Yorkshire 9.1 19.7
Rest of Great Britain 5.3 11.3
Source: KPMG (2013).
C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22 17
variables for which benefits exceed costs. Evidence from such studies
is summarised in the next section. The second is to examine
actual ex post case studies. The following sections do that.
3.2. The modelling approach
De Rus and Nombela (2007) and de Rus and Nash (2009) have
explored the key parameters determining the social viability of
high speed rail, and in particular the breakeven volume of traffic
under alternative scenarios. They built a simple model to compute
capital costs, operating costs and value of time savings for a new
self contained 500 km line at different traffic volumes. Typical
costs were estimated using the database compiled by UIC
(Table 2). A range of time savings from half an hour to one and a
half hours was taken, and a range of average values of time from
15 to 30 euros per hour. Other key assumptions are the proportion
of traffic that is generated, and the rate of traffic growth.
Table 9 shows the breakeven volume in terms of millions of
passengers per annum in the first year, assuming all travel the full
length of the line, under a variety of assumptions about the other
factors. If on average passengers travel half the length of the line,
then of course the required number is doubled. Note that benefit
growth may occur because of rising real values of time as incomes
rise, as well as traffic growth. With exceptionally cheap construction,
a low discount rate of 3%, very valuable time savings and high
values both for the proportion of generated traffic and for benefit
growth, it is possible to find a breakeven volume as low as 3 m
trips per annum, but it is doubtful whether such a favourable combination
of circumstances has ever existed. Construction costs of
30 m euros per km will carry this up to 7 m, and a reduction of
the value of time savings to a more typical level to 4.5 m; lower
benefit growth and levels of generated traffic will take the result
to 4.3 m. An increase in the rate of discount to 5% would take the
value to 4.4 m. In other words, it appears to be the construction
cost that is the key determinant of the breakeven volume of traffic;
all the other adjustments considered have a similar smaller impact.
All of these adjustments together would raise the breakeven volume
to 19.2 m trips per annum, and even worse scenarios can of
course be identified. On the other hand a more modest increase
of capital costs to 20 m euros, with a high value of time savings
but a discount rate of 5%, 30% generated traffic and a 3% annual
growth in benefits leads to a breakeven volume of 9 m. This represents
a realistic breakeven volume for a completely new self contained
high speed line in favourable circumstances. All the
breakeven volumes given assume end-to-end journeys; if some
journeys only use part of the route, breakeven volumes would be
proportionately higher.
These representative breakeven volumes ignore any net environmental
benefits, but we have given reasons above to expect these to
be small. What they also ignore is any network benefits in terms of
reduced congestion on road and air, and also within the rail sector,
and wider economic benefits. If these effects are significant then
HSR may be justified at lower volumes.
3.3. Ex post evaluations
The number of ex post evaluations to be found in the literature
is rather limited, and of course none are truly ex post inasmuch as
they were generally undertaken something like 5 years after opening
of a very long lived asset. So these appraisals still involve forecasting,
but forecasting from the position of knowing what the
construction cost turned out to be and having data for the first
few years of quality of service, traffic, revenue and operating cost.
In France it is a legal requirement that all major government
funded projects are subject to an independent ex post evaluation,
to check that they have provided value for money and to learn lessons
from any problems that are found. Table 10 provides the
results of ex post appraisals of the first six French high speed lines.
Ex post financial and socio economic returns are generally
somewhat below forecast, in some cases due to cost overruns
and in others due to shortfalls of traffic, but the only really large
error was in the case of TGV Nord, which carries traffic to Lille,
Brussels and London, and where traffic through the Channel
Tunnel to London was only a third of that forecast. Although the
only line which might be considered to yield a truly commercial
rate of return is the first one, Paris–Lyon, the returns on all of them
are acceptable in socio economic terms (though only just in the
case of TGV Nord; at the time the French government sought a
minimum return of 5% in cost benefit terms on its investments).
It should be noted that these returns are based on traditional
cost–benefit analysis; that is to say they make no allowance for
any wider economic benefits.
By contrast the experience of Spain has been less satisfactory.
Traditional cost–benefit analysis of the two busiest Spanish lines,
Table 9
Breakeven demand volumes in the first year (m passengers) under varying
assumptions.
Construction
cost (£k per
km)
Rate of
interest
(%)
Value of
time
saved
(euros)
%
generated
traffic (%)
Rate of
benefit
growth
(%)
Breakeven
volume (m
passengers)
12 3 45 50 4 3
12 3 30 50 4 4.5
30 3 45 50 4 7.1
12 3 45 30 3 4.3
12 5 45 50 4 4.4
30 5 30 30 3 19.2
20 5 45 30 3 8.8
Source: Derived by the author from de Rus and Nash (2009).
Table 10
Ex post appraisal of French high speed line construction.
Sud Est Atlantique Nord Inter connection Rhone Alpes Mediterranean
Length (km) 419 291 346 104 115 259
Infrastructure cost Ex ante 1662⁄
2118 2666 1204 1037 4334
(m euros 2003) Ex post 1676 2630 3334 1397 1261 4272
% change +1 +24 +25 +16 +22 À1
Traffic at opening (m pass) Ex ante 14.7 30.3 38.7 25.3 19.3 21.7
Ex post 15.8 26.7 19.2 16.6 18.6 19.2
% change +7.5 À12 À50 À34 À4 À11.5
Financial return (%) Ex ante 15 12 12.9 10.8 10.4 8
Ex post 15 7 2,9 6.5 6.1 4.1
Social return (%) Ex ante 28 23.6 20.3 18.5 15.4 12.2
Ex post 30 12 5 13.8 10.6 8.1
Source: Conseil Général des Pont et Chaussées (2006) Annex 1. Ex post results for the last two lines are taken from Crozet (2014).
18 C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22
Madrid–Seville and Madrid–Barcelona, shows benefits considerably
below costs. (Table 11). The discount rate used was 5%,
but a sensitivity test with 3% gave the same conclusion. It is clear
that the main reason for these disappointing results is the much
lower level of traffic than in France – only 5.5 m passenger trips
p.a. in the first full year on Madrid–Barcelona and 2.8 m. on
Madrid–Seville.
Our final example of an (incomplete) ex post evaluation is the
line from London to the Channel Tunnel in Britain. Originally it
was intended that this line would be built privately on a commercial
basis, but when it became clear that rail traffic through the
Channel Tunnel on Eurostar trains between London and Paris and
Brussels was only reaching around a third of the level forecast,
the private company concerned approached the government for
assistance. At that stage a full cost–benefit analysis was undertaken
(Table 12).
It will be seen that benefits to users form the majority of benefits.
Whilst the biggest share of these were to international traffic,
there were also expected to be substantial benefits to domestic
passengers on commuter services serving a number of south east
towns which would join the high speed line for the run to
London (these operate at 225 kmph on the high speed line rather
than the full 300 kmph of the Eurostar trains). Benefits from
reduced road congestion and reduced environmental externalities
are estimated from the forecast reduction in road traffic and standard
values for different types of road by time of day. The value of
regeneration in the Stratford area was estimated by estimating the
number of new jobs that would be created in the area, and
multiplying this by the amount the government was willing to
pay under other schemes to create jobs in priority areas for
regeneration. This was not at the time a standard part of the
appraisal process.
The costs shown in the appraisal are costs to government,
including grants and government spending.
The (partial) ex post appraisal was undertaken by the National
Audit Office (National Audit Office, 2012). It should be said that
this followed the current approach to appraisal, which is not consistent
with that used in 1998, and included a lower discount rate
and a longer (60 years) assumed life, as well as rising values of time
over time. By this time, on this basis the present value of time savings
had increased to £7b. However, the cost to the government
had increased to £10.2b in present value terms. It made no attempt
to quantify the other benefits of the line, but noted that these
would have to total £8.3b for the line to have a benefit–cost ratio
of 1.5, the level required for a project to be seen as offering medium
value for money in the UK. The main problem was a 30% shortfall
on patronage of international services compared with the
estimate made at the time of the 1998 appraisal (Booz and Co,
2012). It appears that the main reason for this is not a failure of rail
to take the predicted market share between London and Paris and
Brussels, but that whereas the forecast assumed continuing rapid
overall market growth, in practice the market ceased to grow.
Also, the rise of low cost airlines had prevented rail from taking
the predicted market share for longer journeys from other British
cities and/or to cities in Europe beyond Paris and Brussels, for
which rail was not competitive in journey time or fare.
The out-turn construction costs were as in Table 13 and were
within the financial provisions made at the time of the approval
of the project in 1998.
It should be noted that Section 1 comprises 74 km, whereas
Section 2 is only 39 km; however, Section 2 includes a 19 km tunnel
into Central London as well as a 2.5 km tunnel under the River
Thames. Infrastructure UK (2010) compares the cost per km of
building HS1 with five comparable projects elsewhere in Europe
using purchasing power parity exchange rates. The mean construction
cost of the other five was £19 m per km, for HS1 stage 1 it was
£24 m and for stage 2 £94 m. This gives some indication of the very
high cost of Section 2, which included the tunnels noted above as
well as the refitting and extension of St Pancras station in
London and new stations at Stratford and Ebbsfleet.
NAO (2012) concludes that, despite carrying 18.1 m passengers
p.a. (9.7 m international and 8.4 domestic), the time savings to
users were inadequate to justify the capital cost. Its value therefore
depends on regeneration and wider economic benefits. In the original
appraisal, quantification of the regeneration benefits of locating
a high speed rail station in Stratford was based on inadequate
methods; subsequent work has suggested a much greater impact
(Colin Buchanan and Partners/Volterra, 2009), although its magnitude
and whether it really represents a net addition to economic
activity remain controversial.
4. High speed 2
4.1. Introduction
High speed 2 is the proposed high speed line from London to
Birmingham, with branches to Manchester and Leeds and connecting
to the existing main lines to Glasgow and Edinburgh. It is worth
examining in some detail because it has been subject to what is
probably the most extensive and controversial appraisal process
of any high speed line.
The first study of the possibility of building a high speed line
from London to the North was undertaken by Atkins for the
Strategic Rail Authority, the government body then responsible
Table 11
Cost benefit analysis of the Madrid–Seville and Madrid–Barcelona HSR.
CBA of high-speed rail in Spain (billions of 2010 euros)
MadridÀSeville MadridÀBarcelona
Costs 6.8 12.4
Benefits 4.5 7.2
Of which time savings 1.6 2.8
Generated traffic 0.8 1.1
Costs saved on other modes 1.9 2.9
External costs saved 0.2 0.4
Net present value À2.3 À5.3
Source: de Rus (2012).
Table 12
1998 Appraisal of HS1 (£m 1997 NPVs).
User benefits – international services 1800
User benefits – domestic services 1000
Road congestion 30
Environmental benefits 90
Regeneration 500
Total benefit 3420
Costs to government 1990
Net present value 1430
Benefit cost ratio (all benefits) 1.72
Benefit cost ratio (excluding regeneration benefits) 1.5
Source: National Audit Office (2001).
Note: At the current exchange rate (January, 2015) £1 equals 1.3344 euros.
Table 13
Out-turn capital costs of HS1 (£m).
Section 1 construction costs 1919
Section 2 construction costs 3778
Station fit-out 109
New depot 357
Total 6163
Source: NAO (2012).
C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22 19
for rail planning (Atkins, 2002). The objective was to examine
whether there was a case for a high speed line, and if so what route
it should follow, in the light of two main objectives – providing
capacity for the rapid growth in rail traffic underway and expected
to continue, and maximising the sum of time savings. Some 16
high speed rail options were examined comprising different combinations
of the sections of track. Obviously these concentrated
on routes linking the largest cities and relieving the overcrowded
East and West Coast main lines. Also considered were building a
conventional line, upgrading existing lines, road widening and airport
expansion. Extensive sensitivity testing was also undertaken,
including examining the impact of four different long run economic
scenarios, examining rail pricing policy and sensitivity to
costs of other modes of transport.
Table 14 shows the results of the appraisal of two of the options.
Option 1 was a simple London to Birmingham line; Option 8
extended this to Edinburgh and Glasgow via Leeds with a separate
branch to Manchester. Both show an adequate benefit cost ratio,
with the incremental benefit cost ratio on the substantial extra
spending on the larger option also being above 2.
A conventional line could be built for 20% less than HSR, but
would lose £5b of benefits for a cost saving of less than £2b,
whilst upgrading of existing lines appeared even less favourable.
The road widening option appeared worthwhile but of less value
than HSR. Also tested was the issue of timing; it appeared best to
build the full HSR network as soon as possible (by 2016) rather
than deferring it to 2021 or 2026. Charging premium fares
reduced the benefits, although it was noted that the demand
model did not permit testing of more differentiated pricing to
target less elastic market segments (it is a common problem of
rail appraisals that yield management systems are inadequately
modelled, leading to an understatement of revenue or demand
or both).
In terms of what it considered, the Atkins study seems to have
been a model of its kind, considering a wide range of options,
including a road alternative, alternative economic scenarios and
examining both fares policy and timing, which are often neglected
in appraisals. Extensive sensitivity testing was undertaken.
There was no immediate follow up to this study, but in the light
of continued growth in traffic, Network Rail (the infrastructure
manager) undertook its own ‘New Lines’ study in 2009 (Network
Rail, 2009) confirming the case for high speed rail. The response
of the government however was to set up a new company, HS2
Ltd., to look at the case for a high speed line from London to
Birmingham as a first stage of a wider network. After reviewing
the options, HS2 concluded in favour of a Y shaped network along
the lines of Atkins option 8 (the extensions to Manchester and
Leeds forming Phase 2). Trains would continue on conventional
lines to a range of cities including York, Newcastle, Liverpool,
Glasgow and Edinburgh.
4.2. Appraisal of HS2
Table 15 shows the results of the 2013 update of the economic
case for the line. With more than 40 m passengers per annum forecast
to use busiest section of HS2 from 2043, it is not surprising
that the benefits are forecast to greatly exceed the costs.
The principal benefits would again be benefits to transport
users. Wider economic benefits were forecast to exist entirely
because of agglomeration benefits from the release of capacity on
existing lines to improve commuter services into the main cities
(in particular London) rather than because of improved inter city
connectivity. Whilst even the first phase of the line from London
to Birmingham would benefit a large network of origins and destinations,
since it would carry trains to major centres of population
such as Manchester and Glasgow which would complete their
journeys on the conventional network, the full Y shaped network
actually shows a higher BCR than the first phase alone. The reason
is that extending the line to Manchester and in particular connecting
it to Leeds and the North East permits better use of the first
phase of the network giving a better overall result.
Table 16 shows the breakdown of transport benefits. It will be
seen that over half of the benefits are time savings, with business
travel time savings being around 40% of the total. But improved
reliability and reduced crowding are also estimated to be substantial
sources of benefits. Since 69% of the users are estimated to
divert from existing rail lines, benefits to other types of traffic from
the release of capacity on them are substantial. On the other hand
26% of traffic is estimated to be induced and only 4% diverted from
car and 1% from air (DfT, 2013). So any benefits in terms of reduced
externalities on these modes are thought to be small. The reason
for such a low level of diversion is that rail already plays a very
important role between most of the cities served by HS2, and with
rail journey times from London to Manchester and Leeds already
little more than 2 h, air has lost most of its market to rail already.
This is a marked contrast to Spain, where previous rail services
were very slow compared to HSR and rail held a small market
Table 14
Appraisal of options 1 and 8 in the Atkins study.
Option 1 Option 8
Net revenue 4.9 20.6
Non financial benefits 22.7 64.4
Released capacity 2.0 4.8
Total benefits 29.6 89.8
Capital costs 8.6 27.7
Net operating costs 5.7 16.3
Total costs 14.4 44.0
NPV 15.3 45.7
B/C 2.07 2.04
Source: Derived from Atkins (2003) Addendum Table 2.1, with transcription errors
corrected.
Table 15
Appraisal of HS2: present value of costs and benefits over 60 years (£b 2011 prices).
Phase 1 Full network
Oct 2013 Oct 2013
Transport benefits (business) 16,921 40,529
Transport benefits (other) 7673 19,323
Other quantifiable benefits 407 788
Indirect taxes (loss to govt) À1208 À2912
Net transport benefits 23,793 57,727
Wider economic impacts 4341 13,293
Total costs 29,919 62,606
Revenues 13,243 31,111
Net cost to government 16,676 31,495
Benefit cost ratio (Inc. WEIs) 1.7 2.3
Source: DfT (2013).
Table 16
Breakdown of benefits from the proposed HS2 scheme (£b 2011 prices).
Phase 1 Full network
Time savings 17,334 45,679
Crowding benefits 4068 7514
Improved reliability 2624 5496
Car user benefits 568 1162
Total transport user benefits 24,594 59,852
Wider economic impacts 4341 13,293
Other impacts 407 788
Loss to government of indirect tax À1208 À2912
Total 28,134 71,020
Source: DfT (2013).
20 C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22
share. On the Madrid–Seville and Madrid–Barcelona lines nearly
half of the traffic has come from air (de Rus, 2012).
Unlike in the earlier study, no appraisal has been undertaken of
alternative timings for construction of the line. Phase 1 (to
Birmingham) has been planned for completion in 2026 and phase
to in 2033. The reason for this timetable was primarily dictated by
funding and availability of physical resources – the plan was to
follow directly on from completion of the major Crossrail project
in London. However, consideration is now being given to an
accelerated timetable.
The project has been strongly criticised by opponents to the
scheme on a number of grounds. Some argued that the original
appraisal was guilty of extreme optimism in terms both of costs
and demand forecasts (Castles and Parish, 2011). As against that,
it may be argued that the cost estimates were based on experience
elsewhere (HS2, 2012) and that for schemes at an early stage of
development in Britain a large margin for optimism bias (67%) is
now added to the estimates. On the demand side, the case for
the project rests on a continued substantial demand growth until
completion of Phase 2 in 2033. It has been argued that improvements
to telecommunications will reduce the demand for long distance
travel, although to date it seems these may have increased
demand for rail relative to other modes because of the greater
possibility to use them whilst travelling compared with road and
air (Le Vine and Jones, 2012). The British government also takes a
cautious approach in assuming that demand growth will cease in
2036, which is only 3 years after opening of the complete scheme;
if growth continued the benefit cost ratio would be greater than
shown in Table 15. In the latest appraisal, an extensive risk analysis
has been undertaken, demonstrating a low probability of the benefit
cost ratio being less than 1.5 (DfT, 2013). Criticisms of the value
of business travel time and of the wider economic benefits predicted
for the scheme have been discussed above (it should be
noted that the KPMG (2013) estimate of an increase in GDP of
£15b p.a. is not included in the official appraisal in Table 15, where
the much lower value of wider economic benefits relates only to
the improvement of commuter services in the conurbations).
Secondly, there has been much criticism of the detail of the
route and of its impact on the existing rail network. At the
London end, it has been argued that it would be better to bring it
into St Pancras and connect directly with the high speed line to
the Channel Tunnel. The station locations at Nottingham/Derby
and Sheffield are not at the city centres, whilst because of lack of
spare capacity at existing stations and difficulties in extending
them, new stations will be built at Birmingham and Leeds, leading
to less than ideal interchange with existing rail services and the
consequent need for investment in local transport, including light
rail, to improve local connectivity. Wellings (2013) argues that
the cost of these measures should be regarded as part of the cost
of HS2, although presumably they add benefits as well. There are
also fears that some cities on the existing main line and not served
by HS2 will incur a worsening of services, although others will gain
from use of the capacity released on those lines. Optimal design of
the route requires numerous options regarding all of these issues
to be examined, and indeed the siting of the Nottingham/Derby
and Leeds stations is being reviewed.
Thirdly has been the claim that other cheaper alternatives to
HS2 have not been adequately examined. In particular, a group of
local authorities opposed to the line (the 51 M group) put forward
a much cheaper package of alternatives to increase capacity at the
London end of the line, costing only £1.2b (Castles and Parish,
2011). However, it would not provide sufficient capacity to meet
peak demand, some of which would have to be dealt with by other
measures such as large increases in peak fares. Atkins were commissioned
to appraise this package (Table 17) and found that the
package would yield a benefit–cost ratio of no less than 6.
However, if we look at the incremental costs and benefits of HS2,
it shows an incremental benefit–cost ratio of around 2, roughly
the threshold for the British government to consider a project to
give high value for money. Thus the additional cost of HS2 is
justified by additional benefits, according to the appraisal.
The latest strategic case published by DfT (2013) includes a
much more thorough examination of alternative rail schemes
(diversion of traffic to road or air is dismissed as incompatible with
government policy). Several different packages of schemes, including
duplicating sections of main line and improving junctions and
existing stations were put together. Table 18 shows the appraisal
of the packages of schemes most comparable to HS2. Neither package
provides nearly as much capacity as HS2. Both provide higher
BCRs than HS2, but the incremental BCR from the much greater
investment in HS2 remains above 2.
Thus the case for HS2 has been subjected to much more extensive
scrutiny than many transport projects, with extensive examination
of alternatives. Nevertheless, such is the scale and
complexity of the project that arguments continue as to whether
all the options and their implications for the network as a whole
and in terms of local transport connectivity have been adequately
examined.
5. Conclusions
From the number of studies undertaken and the experience of
high speed lines around the world, it is possible to reach some general
conclusions on the circumstances in which high speed rail will
be worthwhile. Of course, the decision is a trade-off between costs
and benefits; factors which make for higher costs may be outweighed
by other factors making for higher benefits. The dominant
factors are construction costs, value of time savings per passenger
and the volume of passengers.
Firstly, high speed line construction costs vary greatly, but a
major factor is the amount of tunnelling. If routes can be found
which allow access to city centre stations without tunnelling the
savings will be large. This might be achieved by placing suburban
services underground; this may be a lower cost solution and have
other benefits if it enables suburban services to penetrate and cross
the city centre.
Secondly, the value of time savings per passenger will vary with
the quality of the alternative and the incomes (and therefore value
of time) of passengers. HSR will bring greater time savings per passenger
where it is substituting for conventional rail or car than
where it depends for most of its traffic on a marginal time saving
compared with air. Of course other benefits – environmental and
Table 17
Incremental benefits and costs over the 51 M alternative package (£b 2011PV).
51 M Y shaped increment
Benefits 7.108 46–52
Costs to government 1.173 25–23
BCR 6.06 1.6–2.3
Source: Derived from Atkins (2013).
Table 18
Capital costs and benefits of alternatives.
HS2
Phase 1
Phase 1
alternative
HS2 both
phases
Phase 1 and 2
alternative
Capital costs (£b) 19.4 2.5 38.4 19.2
Benefit–cost ratio 1.7 2.0 2.3 3.1
Benefits (£b) 28.1 8.5 71.0 30.7
Source: DfT (2013).
C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22 21
reduced congestion at airports – may be greater where it is predominantly
diverting from air, but it is unlikely that these will
compensate for the smaller time savings in the appraisal.
Thirdly, serving a large population is crucial. High speed rail
requires either to link very large cities, or to serve a string of large
cities, possibly by running on conventional lines past the end of the
high speed line. High density cities with strong public transport
networks will favour HSR over car and air. HSR journey times of
around 3 h are required to compete effectively with air, however,
so HSR will be most effective for routes of up to 700 km. For shorter
journeys HSR may be worthwhile in terms of time savings to
existing rail users and diversion from car, but below around
150 km, the high speed of HSR will be of less importance because
of the shortness of the journey.
Fourthly, congestion on existing rail, road and air systems will
favour HSR by providing a case for more capacity, not just for fast
intercity trains but also for commuter and regional passenger
trains and freight, either on the high speed line (as with commuter
trains on HS1) or in the capacity released on existing lines.
However, as demonstrated by the British example, there are
always numerous ways of providing more capacity or of using
demand management to remove the need for it; fitting a new high
speed line into an intensely used existing network is a complex
matter requiring many options to be examined (including timing).
A policy which favours rail over road and air in terms of adding
capacity on environmental grounds will obviously favour the case
for HSR.
Finally the most controversial and unresolved issue is the
extent to which high speed rail will produce wider economic benefits.
Were such benefits to be firmly identified they might reduce
the level of traffic needed to justify the investment by a substantial
amount.
Acknowledgements
An earlier version of this paper was produced as Christopher
Nash (2013) When to Invest in High-Speed Rail. International
Transport Forum Discussion Paper No. 2013-25 December. OECD,
Paris.
I am grateful to ITF for permission to reuse material from that
paper, and to participants in the round table on High Speed Rail
in New Delhi, December 2013 for comments on the earlier draft.
References
Albalate, D., Bel, G., 2012. The Economics and Politics of High Speed Rail. Lexington
Books, Maryland.
Atkins, 2002. High Speed Line Study, HSL Business Case. Atkins, London.
Atkins, 2003. High Speed Line Study, Summary Report. Atkins, London.
Atkins, 2013. HS2 Strategic Alternatives, Final Report. DfT, London.
Batley, Richard, Mackie, Peter, Wardman, Mark, 2013. Review of the Value of Time
Assumptions for Business Travellers on HS2 DfT, London.
Booz, Co, 2012. Review of HS1 Demand Forecasts, Report to HS2 Ltd. HS2 Ltd.,
London.
Castles, C., Parish, D., 2011. Economic Evaluation: London – West Midlands Link.
RAC Foundation, London.
Colin Buchanan and Partners/Volterra, 2009. Economic Impact of High Speed 1,
Final Report. London & Continental Railways, London.
Campos, J., de Rus, G., Barron, I., 2009. A review of HSR experiences around the
world. In: de Rus, G. (Ed.), Economic Analysis of High Speed Rail in Europe.
Fundacion BBVA, Bilbao (Chapter 1).
Chartered Institute of Logistics and Transport, 2011. Transport Use of Carbon. CILT,
London.
Civity, 2013. Further Development of the European High Speed Rail Network
System Economic Evaluation of Development Options, Summary Report. Study
commissioned by Alstom and SNCF, Paris/Hamburg.
Crozet, Y., 2014. High-Speed Rail Performance in France. From Appraisal
Methodologies to Ex-post Evaluation. International Transport Forum
Discussion Paper No 2013-26 OECD, Paris.
de Rus, Nash, C.A., 2009. In what circumstances is investment in HSR worthwhile?
In: de Rus, G. et al. (Eds.), Economic Analysis of High Speed Rail in Europe.
Fundacion BBVA, Bilbao, chapter 3.
de Rus, G., Nombela, G., 2007. Is investment in high speed rail socially profitable? J.
Transp. Econ. Policy 41 (1), 3–23.
de Rus, G., 2012. Economic Evaluation of the High Speed Rail, undertaken for The
Expert Group on Environmental Studies. Ministry of Finance, Sweden, <http://
www.ems.expertgrupp.se/Default.aspx?pageID=3>.
Department for Transport, 2013. HS2 Strategic Case. DfT, London.
GRACE, 2005. Generalisation of Research on Accounts and Cost Estimation,
European Commission project under the Transport RTD of the 7th Framework
Programme, Deliverable D7. University of Leeds, Institute for Transport Studies.
Graham, D.J., 2007. Agglomeration, productivity and transport investment. J.
Transport Econ. Policy 41 (3), 317–343.
Graham, D.J., Melo, P., 2010. Advice on the Assessment of Wider Economic Impacts:
a Report for HS2. HS2 Ltd., London.
Greengauge, 2012, The Carbon Impacts of HS2. London.
Hensher, D.A., 1977. Value of Business Travel Time. Pergamon Press, Oxford.
HS2 Ltd, 2012. Cost and Risk Model. HS2 Ltd., London.
INFRAS, CE Delft, Fraunhofer Gesellschaft ISI, University of Gdansk, 2008. Handbook
on Estimation of External Cost in the Transport Sector, produced within the
study Internalisation Measures and Policies for All External Costs of Transport
(IMPACT). CE Delft, Delft.
Infrastructure UK, 2010. Infrastructure Cost Review. London.
KPMG, 2013. High Speed Rail: Consequences for employment and economic
growth, Technical Report. HS2 Ltd., London.
Kurosaki, F., 2014. Shinkansen Investment before and after JNR Reform,
International Transport Forum Discussion Paper No. 2013–2027. OECD, Paris.
Le Vine, S., Jones, P., 2012. On the Move: Making Sense of Car and Train Travel
Trends in Britain. University College London, London.
Lyons, G., Jain, J., Holley, D., 2007. The use of travel time by rail passengers in Great
Britain. Transp. Res. Part A 41, 107–120.
Marks P., Fowkes A.S., Nash C.A., 1986. Valuing Long Distance Business Travel Time
Savings for Evaluation: A Methodological Review and Application. PTRC
Summer Annual Meeting.
Mohring, H., Williamson, H.F., 1969. Scale and ‘‘industrial reorganisation’’
economies of transport investments. J. Transport Econ. Policy 3 (3), 251–271.
National Audit Office, 2001. Department of the Environment, Transport and the
Regions: The Channel Tunnel Rail Link. HMSO, London.
National Audit Office, 2012. Department for Transport: The Completion and Sale of
High Speed 1. London.
Network Rail, 2009. New Lines Programme. Strategic Business Case, London.
Overman, H., 2013. Treasury Committee, Oral evidence: Economics of HS2, HC 788
Tuesday 5 November 2013.
Paix, J.-F., 2010. Feedback on High Speed Rail in France, Proceedings of the 7th
World Congress on High Speed Rail, Beijing, UIC.
Bridget Rosewell, Tony Venables, 2013. High Speed Rail, Transport Investment and
Economic Impact. Paper written for HS2 Ltd., London.
Sanchez-Borras, M., 2010. High Speed Rail in Spain, presentation at the first TEMPO
Conference on Sustainable Transport, Oslo.
SDG, 2004. High Speed Rail: International Comparisons, Final Report. Commission
for Integrated Transport, London.
SDG, 2006. Air and Rail Competition and Complementarity, Final Report. European
Commission, DGTREN, London.
Venables, A.J., 2007. Evaluating urban transport improvements: cost–benefit
analysis in the presence of agglomeration and income taxation. J. Transport
Econ. Policy 41 (2), 173–188.
Wardman, M., 2004. Public transport values of time. Trans. Policy 11, 363–377.
Wellings, R., 2013. The High-Speed Gravy Train, IEA Current Controversies Paper No.
46, London.
Wu, Jianhong, Nash, Chris, Wang, Dong, 2014. Is high speed rail an appropriate
solution to China’s rail capacity problems? J. Transp. Geogr. 40 (2014), 100–111.
22 C. Nash / Journal of Rail Transport Planning & Management 5 (2015) 12–22