Memorandum by Dr Denvil Coombe (RP 49)
ROAD PRICING
CREDENTIALS
Led the South and West Yorkshire and London
Orbital Multi-Modal Studies, both of which recommended that area-wide
road pricing should be introduced as soon as practical as the
only effective means of tackling congestion in the medium to longer
term. The teams also recommended that limited motorway widening
should be undertaken and that, in the meantime, the benefits from
this widening should be "locked in" by means of management
measures integral to the widening schemes.
Led the A3 Hindhead Roads-based Study in which
the case for tolling the proposed tunnel under Hindhead Common
was investigated. The study team recommended against charging
for the use of the tunnel alone because too much traffic would
avoid payment of the tolls by diverting to alternative untolled
routes and advised that tolls should be reconsidered only when
the technology for free-flow tolling became available and only
in the context of an area-wide road pricing scheme.
Contributed to the Department for Transport's
Feasibility Study of Road Pricing in the UK by (a) preparing a
review of the Multi-Modal Study and Roads-based Study evidence
on road pricing; (b) assisting the South and West Yorkshire Multi-Modal
Study consultants in the conduct of tests of alternative road
pricing strategies; and (c) preparing a Composite Report of the
tests of alternative road pricing strategies conducted using models
covering the areas studied in the South and West Yorkshire, London
Orbital and Cambridge to Huntingdon Multi-Modal Studies.
Member of the Standing Advisory Committee on
Trunk Road Assessment for the 1994 Report on Trunk Roads and the
Generation of Traffic and the 1999 Report on Transport and the
Economy.
SOME COMMENTS
ABOUT THE
PROPOSED M6 EXPRESSWAY
The Department for Transport set out their proposal
for a new dual two-lane, tolled, motorway alongside the existing
M6, from Junction 11a to Junction 19, in their document M6: giving
motorists the choice, dated July 2004. The main intention
is to provide more road capacity in the corridor, at a lower cost
compared with widening the existing M6. Applying tolls on the
new road would enable traffic levels on the new road to be controlled
so that an attractive level of service can be maintained, as well
providing a revenue stream to assist with its funding.
In the short term, the main response from drivers
is that some of them would re-route to take advantage of the new
capacity. Leaving all other responses on one side, I expect average
traffic speeds would rise as a result of re-routeing alone.
Another short-term response is likely to be
that some drivers would re-time their journeys to take advantage
of the reduced congestion at peak times. This would result in
higher speeds in the periods outside the peaks and would depress
speeds somewhat in the peak periods back towards the pre-improvement
levels.
In the longer term, but not necessarily very
long term, I would expect some drivers to adjust their destinations
to take advantage of the fact that they can travel further in
any given period of time than they could before the new road was
open. Some changes of this nature—for example, changes in
trips for discretionary purposes, such as leisure and shopping-may
occur in the shorter term, while other changes in destination,
such as changes in workplaces, may take longer to mature.
In the somewhat longer term, I would expect
some people to respond by moving home, again to take advantage
of the fact that they can travel further in any given period of
time.
There may also be some transfer from public
transport and some generation of wholly new motorised trips, although
the evidence available suggests that, in general, these effects
will be smaller than the trip-lengthening effect caused by changing
where people live, work and undertake other activities.
Also in the longer term, but again not necessarily
in the very long term, I would expect pressures to arise for development
at locations close to either the junctions on the new Expressway
or junctions on the existing M6 which would be substantially relieved
by the new road. These developments would, of course, give rise
to extra traffic.
The net effect of all these factors could well
be an increase in traffic in the corridor, over and above the
increase that would have occurred without the new road. Without
seeing the traffic model output, the scale of the extra traffic
which would be induced by the new capacity is hard to judge. Experience
suggests that the scale would be higher where congestion in the
pre-improvement case is high. An increase in the road capacity
in the M6 corridor is being seen as a priority simply because
it is congested. In general terms, the greater the degree of congestion
prior to an improvement, the greater will be the amount of extra
traffic induced by the improvement. I would therefore expect the
level of induced traffic to be relatively high in this case.
The answer to the problems caused by induced
traffic is to implement measures on the existing M6, and possibly
on other major roads in the corridor as well, to "lock-in"
the benefits of the extra capacity.
The best way of achieving this, in the absence
of an area-wide road pricing scheme, would be to introduce tolls
on the existing M6 as well as on the new Expressway. The idea
is that the tolls on the existing M6 would be set at a low level,
just sufficient to deter induced traffic and thereby maintain
total traffic in the corridor at the pre-improvement levels. Thus,
all drivers using the existing M6 and the new M6 Expressway would
pay, but those wishing to pay for the higher level of service
offered by the M6 Expressway would pay more. If this is not feasible,
then a combination of ramp metering and high occupancy vehicle
lanes may be deployed in an attempt to contain total traffic levels
in the corridor as nearly as possible to their pre-improvement
levels.
M6: giving motorists a choice says in
paragraph 13: "In July 2003, the Secretary of State for Transport
published the discussion paper Managing Our Roads, which
outlined the key issues involved in tackling congestion on the
strategic road network. These include getting the right balance
between additional capacity and measures which ensure benefits
are `locked in', whether through physical measures or pricing.".
While this acknowledges the point about the need to implement
demand management in conjunction with providing new capacity,
there is no discussion of what this might mean in the context
of the proposed M6 Expressway.
The recommendation from the West Midlands to
North West Multi-Modal Study was for an additional lane in each
direction on the existing M6. The proposal for an M6 Expressway
effectively doubles the extra capacity that is to be provided.
Thus, if the additional one lane in each direction was an appropriate
level of increase, the new M6 Expressway must represent an over-provision
of capacity. If this is the case, the new M6 Expressway would
tend to undermine rather than support the case for a national
road pricing scheme in that the justification for a road pricing
scheme is dependent on congestion levels being sufficiently high
and, if congestion has been reduced unduly by the over-provision
of road capacity, the justification for road pricing would be
reduced.
SOME OBSERVATIONS
ABOUT AREA-WIDE
ROAD PRICING
The basic principle of efficient road user charging
is that users should, in theory, be charged for all the additional
costs they impose by their use of the road system—road wear
and tear, delays to other users, increased accident risk and environmental
cost. The term used to refer to these additional costs is "Marginal
External Cost". When these are added to those costs borne
directly by the user (such as fuel and their own time), the result
is called "Marginal Social Cost".
In the Feasibility Study of Road Pricing in
the UK, published by the Department for Transport in July 2004,
patterns and levels of charges were sought, using the Department's
National Transport Model, which closely matched Marginal Social
Costs. The result was a set of charge rates which varied by area
type, road type and volume to capacity ratio and which ranged
from small negative to large positive charges per kilometre.
This set of charges was considered as being
both additional to fuel duty and a replacement for part of fuel
duty such that the total revenue remained the same as in the case
without charging. The comments which now follow relate to the
case where road user charges are additional to existing fuel duty
and other motoring taxes.
While the approach of applying charge rates
which closely match Marginal Social Costs is theoretically correct,
it can be problematic in practice. In reality, Marginal Social
Costs will vary from link to link and finding a charging structure
which (a) matches closely the pattern of actual Marginal Social
Costs, (b) is comprehensible to road users, and (c) does not cause
unwanted re-routeing effects, can be very difficult at least and
impossible in some circumstances.
The pattern of 80 charge rates developed by
the Department for its recent study was aimed at matching Marginal
Social Costs reasonably closely but actually yielded unwanted
re-routeing effects. When the charge rates devised using the National
Transport Model were passed down to the Multi-Modal Study models
it was found that the high degree of variability in charge rates
from link to link caused circuitous routeing, to the extent that
the average trip length increased rather than decreased as expected.
The result was that, whereas the National Transport Model appraisal
showed a net benefit from Marginal Social Cost pricing, the Multi-Modal
Study model in which the trip-lengthening effect was most clearly
evident showed a net disbenefit because the effects of the extra
traffic outweighed the gains from fewer trips using the road system.
A much more approximate approach was used in
the South and West Yorkshire and London Orbital Multi-Modal Studies
in which a single figure area-wide distance-based charge was sought
which maximised user benefits. Simple charging systems of this
kind are quite distant from true Marginal Social Cost pricing
in which charges may well vary markedly from link to link.
In practice, a system which balances the aims
of matching Marginal Social Cost, maintaining comprehensibility
for users, and avoiding perverse and/or unwanted effects is likely
to lie somewhere between a true Marginal Social Cost pricing system
(in which charge rates are likely to be highly variable from link
to link) and a single figure area-wide distance-based charge.
The following distinction may be made between
Short Run and Long Run Marginal Social Costs. Short Run Marginal
Social Cost is defined as the social cost of an additional vehicle
kilometre at the current level of infrastructure provision. Long
Run Marginal Social Cost is defined as the social cost of an additional
vehicle kilometre when infrastructure provision is optimally adjusted
to the level of demand.
For simplicity, I consider now the components
of Short Run Marginal Social Cost only: road maintenance and renewal
costs (borne by infrastructure providers); congestion and accidents
(borne by infrastructure users); and noise and local air pollution
and global warming (borne by non-infrastructure users). Again,
for simplicity, I consider only congestion, which has been at
the heart of the case for road pricing since first conceived in
the 1920s.
When charges are levied for the use of congested
road infrastructure, the two main groups of response will be as
follows: some road users will elect not to pay the charge but
will instead either not travel at all or will travel by another
mode; and other road users (usually the majority) will decide
to pay the charge and continue to travel by road, although some
may change their destination to one nearer their origin and many
may change their route. By "road users" here, I mean
those that use private vehicles of some kind (cars, motorcycles
and goods vehicles) and I exclude public transport passengers.
These two groups of response will occur with all charging systems,
no matter how closely the charging system replicates the actual
pattern of Marginal Social Costs.
The people in the first group will all suffer
a disbenefit. Those who do not travel at all will lose the utility
of the journey they made before the pricing was introduced. Those
who travel by an alternative mode will also suffer a disbenefit
because, as far as they are concerned, their total generalised
cost, without the charge but including their valuation of travel
quality and convenience, will have been less by private vehicle
than it will be by the alternative mode.
Leaving aside for the moment the impacts of
the charges (which I deal with later), most of the people in the
second group will experience a benefit from reduced congestion
and reduced vehicle operating costs (especially if their routes
are less circuitous because of the reduced congestion) caused
by fewer vehicle-kilometres being travelled on the road system.
However, those who respond by travelling to a closer destination
(as a result of the charge) will lose the utility derived from
travelling to their previous destination further away. It is not
possible to disentangle these effects but, for a road pricing
scheme to be justified, the net time and vehicle operating cost
benefits to this group of travellers as a whole will usually be
positive.
For road pricing to be justified as a congestion
reducing measure, the net benefits to the second group should
outweigh the disbenefits to the first group, to yield an overall
net benefit in these terms. This will not happen where: congestion
is low and the benefits from reductions in vehicle-kilometres
travelled on the road system are small; and/or charges are set
so high as to deter too many people from travelling by private
vehicle thereby generating high disbenefits; and/or the alternative
modes are very poor compared with the private vehicle, again thereby
generating high disbenefits. So the conditions have to be right
and it is not the case that road pricing can be justified in all
cases.
A further factor to consider now is the charges.
Note that I am only considering the case where the charges are
additional to existing motoring taxes. These are a disbenefit
to those who pay them—the continuing road users—and
a benefit to the charging authority. In very broad terms, these
two terms cancel out in the cost/benefit analysis, but not precisely
because of the way the calculations are done. There will also
be changes in tax revenues to Government but these are not usually
so large as to distort the simplified story I am setting out here.
The other key element of the cost/benefit analysis will be the
costs of collecting the revenues, including the costs of setting
up the system and operating, maintaining and enforcing it.
It is generally the case that the charges paid
by continuing road users will outweigh the net benefits to travellers
as a whole. So leaving aside taxation changes, and the complications
of the way that the calculations are performed for each of the
benefits and disbenefits, the general picture can be summarised
in the following simple terms:
— net benefits (in terms of time and
vehicle operating cost savings) to continuing road users (+A);
— disbenefits to those deterred from
travelling by private vehicle (and who either do not travel or
travel by a less suitable mode) (-B);
— charges paid by continuing road users
(-C);
— revenues accruing to the charging
authority (+D); and
— costs of collecting the charges (-E).
For a scheme to be justified as a congestion
reducing measure, +A-B will be positive. Usually, the magnitude
of C will exceed the magnitude of +A-B, so that +A-B-C will be
negative. D will broadly cancel out C. And +A-B will exceed E
(summed over the appraisal period). Usually, the net revenues
(+D-E) will be very large. Substantial further benefits can be
gained, in principle, from investing (or recycling) these revenues
in other good value for money projects, including schemes to compensate
some of the adversely impacted groups.
Some commentators would argue that, under a
good road pricing scheme, it is possible, in theory, to ensure
that everyone benefits. This statement is true if +A-B-C+D-E is
positive and if, somehow, the net revenues can be used to compensate
losers directly. In practice, of course, this is not possible
and so there will be some losers among the winners.
In the simple terms set out above, travellers
as a group will usually experience a net loss. Within that generality
though, there will be a distribution of impacts, ranging from
winners to losers. The winners will be those with high values
of time on high value trips (such as business)—their gains
from reduced congestion will outweigh the charges they pay. The
losers will be those continuing road users with low values of
time-the charges they pay will outweigh their gains from reduced
congestion—and those who either do not travel or who change
mode. Thus, in general, car travellers on business and possibly
goods vehicles (depending on whether or not they pay significantly
higher charges than cars) will experience a net gain, while other
car travellers will experience a net loss. The sections of society
too poor to own a car will gain from faster bus speeds as a result
of the reduced congestion. They may also gain more substantially
if some of the net revenues are used to provide better public
transport services and other measures aimed at reducing social
exclusion. Those travellers who change mode as a result of the
charge may also gain if their new mode can be improved by use
of some of the net revenues from the charging. Those most difficult
to compensate are the low income car users who continue to travel
by car and pay the charge, although the recycling of revenues
to improve public transport may, if targeted carefully, partially
offset the disbenefits to this group.
The impacts set out above will change as the
charge level increases in different ways. In general terms, the
shapes of the relationships between the impacts and the charge
levels will be as follows:
— the aggregate net time and vehicle
operating cost benefits to continuing road users (A above) will
increase but the rate of increase will tail off as traffic levels
are reduced due to the nature of the flow/delay or speed/flow
relationships (or, in other words, the shape of the supply curve),
as shown below (the vertical axis is the net time and vehicle
operating cost benefit and the horizontal axis is the charge);
— the aggregate disbenefits to those
deterred from travelling by private vehicle (B above) will increase
but the rate of increase will accelerate as the charges impact
on those with less attractive alternatives, due to the shape of
the demand curve, as shown below (the vertical axis is the disbenefit
to deterred travellers (of opposite sign to the benefits in the
previous point) and the horizontal axis is the charge);
— the disbenefits of the charges to
road users (C above) will increase up to a maximum and then decline
as the charges reach levels where the levels of deterrence become
high (this is a function of the shape of the demand curve), as
shown below (the vertical axis is the disbenefit of the charges
to continuing users and the horizontal axis is the charge);
— the revenues accruing to the charging
authority (D above) will increase up to a maximum and then decline
to mirror broadly the relationship between the charge disbenefit
and the charges shown in the previous point, as shown below (the
vertical axis is the benefit of the revenues to the charging authority
and the horizontal axis is the charge);
— the costs of collecting the charges
should be broadly unaffected by changes in charge level.
As noted above, the charge disbenefits and the
revenue benefits broadly cancel out and it should be possible
to see that the combination of the first two graphs should give
a maximum benefit. It follows that the ease with which a maximum
can be determined will depend on:
— the shape of the supply curve (that
is, the flow/delay or speed/flow relationship) for the network
as a whole and the slope of that curve at the level of demand
on the network; and
— the elasticity of demand or the slope
of the demand curve in the vicinity of its intersection point
with the supply curve (that is, the point of equilibrium between
supply and demand).
It is not a foregone conclusion that a maximum
can be determined in every network. If the demand is too low and
the equilibrium point is on the flat part of the supply curve,
or the demand curve is too flat, it may be difficult to identify
a clear maximum.
In principle, the single charge level that will
yield the maximum economic benefits can generally be identified
quite easily, providing the circumstances are right. By "single"
charge level, I mean, for example, either a single area-wide distance-based
charge or a particular level of a single pattern of cordon crossing
charges. While it may be sufficient for some purposes to define
the optimum charge in these simple terms, there are, in principle,
many variations which could be explored in search of higher maximum
economic benefits. In general terms, the variations should be
with the aim of matching more closely the distribution of Marginal
Social Costs throughout the network, bearing in mind the need
to maintain comprehensibility and avoid perverse and/or unwanted
effects.
In the South and West Yorkshire Multi-Modal
Study, the economically optimum single figure area-wide distance-based
charge was shown to be 4 pence/km (additional to existing motoring
taxes). Reducing the charge to 2 pence/km for the links where
the volume to capacity ratio was less than 25% and increasing
the charge to 7 pence/km where the volume to capacity ratio exceeded
75% increased the benefits by about 40%. While it was clear from
this test that the value of the maximum benefits can be affected
by the charging structure, no tests were undertaken to assess
whether the general charge level at which the maximum benefits
occurred would change.
The flow reductions that would arise with the
economically optimum charging level will vary with the circumstances—essentially,
with the slopes of the supply and demand curves. Even with a single
area-wide optimum charge, flow reductions will vary from link
to link as traffic adjusts to the combination of lower traffic
levels in combination with the charge. Variations in charge levels
from link to link will also cause flow reductions to vary.
Turning now to the other components of Short
Run Marginal Social Cost, I imagine that road maintenance and
renewal costs will decrease with reducing traffic. Accidents are
related to traffic volumes and accidents per vehicle-km vary by
road type. Road user charging may alter the distribution of vehicle-kms
by road type and so the relationship between reduced traffic as
a result of a road user charging scheme and reduced accidents
is not necessarily a direct one although it may well be broadly
linear. Noise levels are a function of traffic flow, proportion
of heavy vehicles and traffic speed and only matter where sensitive
receptors exist. The relationship between traffic reductions arising
from road user charging and the numbers of people annoyed by noise
is therefore very complex. Air pollutants are a non-linear function
of traffic flows, proportion of diesels and traffic speed. The
relationship between traffic reductions and changes in local air
quality are therefore very hard to pre-judge without actually
performing the calculations. Greenhouse gas emissions, however,
are directly related to vehicle-kilometres and are thus likely
to decrease in line with reductions in traffic.
From this simplistic analysis, it can be seen
that the effects on the definition of the optimum road user charge
of including all the elements of Short Run Marginal Social Costs
are very unclear. However, in the Department's tests undertaken
using the National Transport Model, simplified calculations of
accident and environmental costs were included. On the basis of
these tests, the teams carrying out the tests using the Multi-Modal
Study models were advised that they could concentrate on time
and vehicle operating cost savings only as these accounted for
approaching 90% of the total benefits.
Finally, turning to the question of the Long
Run Social Marginal Cost, determining the optimum level of charging
in this case is more complicated because, in principle, the capacity
of the road system should be adjusted to cater for the demand
from each optimum charging structure and level. The addition of
new infrastructure will, in turn, affect the position of the optimum
benefits, and so some iteration is likely to be required to determine
the optimum combination of charge level and structure and amount
of new capacity.
KEY MESSAGES
Toll Roads
If toll roads are to be considered, two principles
should apply. First, the net increase in capacity should not exceed
that required under an optimum area-wide road pricing scheme in
the first year that such a scheme would be justified and practical.
This would be consistent with the statement on page 7 of The Government's
Response to the Transport Select Committee's Report Jam Tomorrow?:
The Multi-Modal Study Investment Plans: "We will ensure
that the case for road schemes taken forward following the Multi-Modal
Studies is robust even if a decision is taken to introduce road
user charging in the future.". Second, the benefits accruing
to the users of roads relieved by the new toll road, including
those on the existing motorway, should be "locked in"
by either tolls on those roads (as well as the new road) or other
management measures.
Area-Wide Road Pricing
The basic principle of efficient road user charging
is that users should, in theory, be charged for all the additional
costs they impose by their use of the road system—road wear
and tear, delays to other users, increased accident risk and environmental
cost. When these costs are added to those borne directly by the
user (such as fuel and their own time), the result is called "Marginal
Social Cost". While the approach of applying charge rates
which closely match Marginal Social Costs is theoretically correct,
it can be problematic in practice. In reality, Marginal Social
Costs will vary from link to link and finding a charging structure
which (a) matches closely the pattern of actual Marginal Social
Costs, (b) is comprehensible to road users, and (c) does not cause
unwanted re-routeing effects, can be very difficult at least and
impossible in some circumstances.
Travellers as a group will usually experience
a net loss from a road pricing scheme. Within that generality
though, there will be a distribution of impacts, ranging from
winners to losers. The winners will be those with high values
of time on high value trips (such as business)—their gains
from reduced congestion will outweigh the charges they pay. The
losers will be those continuing road users with low values of
time—the charges they pay will outweigh their gains from
reduced congestion—and those who either do not travel or
who change mode. Thus, in general, car travellers on business
and possibly goods vehicles (depending on whether or not they
pay significantly higher charges than cars) will experience a
net gain, while other car travellers will experience a net loss.
The sections of society too poor to own a car will gain from faster
bus speeds as a result of the reduced congestion. They may also
gain more substantially if some of the net revenues are used to
provide better public transport services and other measures aimed
at reducing social exclusion. Those travellers who change mode
as a result of the charge may also gain if their new mode can
be improved by use of some of the net revenues from the charging.
Those most difficult to compensate are the low income car users
who continue to travel by car and pay the charge, although the
recycling of revenues to improve public transport may, if targeted
carefully, partially offset the disbenefits to this group.
In principle, there is an optimum combination
of charge level and structure and the amount of new road capacity.
Thus, the case for new road capacity, whether provided by traffic
management or new infrastructure, should be considered in the
context of, and jointly with, area-wide road pricing.
Denvil Coombe
Transport Planning Consultant
January 2005
| 
