Planning and Fuel Use: a
Critical Survey
Alan W. Evans
University of Reading
Or –
Lies, Damned Lies, and Statistics
• With apologies to Mark Twain and/or
Benjamin Disraeli
Introduction
• Land Use Planning, Fuel Use and CO2
Emissions
• Planning Deals with Externalities
• CO2 Emission is an Externality
• Therefore Planning can Deal with it
Dr Pangloss Finds His Profession
• ‘High densities result in reduced car use,
therefore we need increased densities.
• But, thankfully, we have green belts and
countryside protection so all is for the best
in the best of all possible (planning)
worlds.
• We just have to ensure that building is at
high density on brownfield sites.’
• All the evidence that is needed is
contained in a figure in Newman and
Kenworthy (1989) showing ‘gasoline use
per capita versus urban density’ for cities
around the world in 1980.
• Suspicious economists might think that this
pattern results from price and income
differences rather than differences in density.
• Newman & Kenworthy have an answer –
another graph which shows the same
relationship ‘adjusted to US income, vehicle
efficiencies and gasoline prices’ in 1980.
• In this format the figure was used to justify
high densities in England by:
• The Rogers Report, 1999
• Planning Policy Guidance, 2000
and, but suspiciously, in
The Barker Report, 2006
• But an even more suspicious economist
might ask how the adjustment is done.
• So London’s estimated gasoline
consumption per capita increases by
about 20% with this adjustment.
• From 12,426 MJ to about 15,000 MJ (the
exact figure is not given).
• But a Londoner’s average income
increases from $4,990 to $8,089 (the
average for US cities) i.e. by 60%
• And fuel prices drop from 70 cents per litre
to 23.6 cents per litre, i.e. by two thirds.
• Maybe it’s because of a fall in fuel
efficiency from 10.7 litres per 100km to the
US average of 15.35 litres per 100km, i.e.
by 43% .
• But that doesn’t make sense either,
particularly as N&K state that in the long
run the question of fuel efficiency would
become unimportant.
• Their view seems to be that in the short
run the sort vehicles which already exist
are important.
• But in the long run people will buy
more/less efficient cars as fuel prices
rise/fall.
• Generally N & K give only averages for
groups of cities – US, Australian,
European, and Asian
• They give figures for only one city, the
Canadian city of Toronto
Toronto
• The only city for which full data is given.
• There incomes would rise from $7,521 to
$8,089, i.e. by 7%
• Fuel prices would change from 23.5c to
23.6c per litre, i.e. by less than ½%
• Fuel efficiency would improve from 16.3 to
15.35 litres per 100km, i.e. by 6%.
• Acording to Newman and Kenworthy these
small changes should result in substantial
falls in consumption per capita.
• From 34,813MJ in 1980
• To 29,995MJ in the short run
• And to 26,090MJ in the long run
• A fall of 14% in the short run and of 25%
the long run!
One Conclusion
• What is shown in the diagram as “adjusted
to US incomes and prices” is calculated
using the ‘short run elasticities’ not the
‘long run’!
• Misleading? Deliberate or accidental?
A Second Conclusion
• There is no way in which an increase in
incomes of 6% can result in a 25% fall in
consumption.
• There is something very wrong with their
figures.
One Explanation
• A research assistant invented the figures
which showed what N&K wanted them to
show so no-one checked them.
• The assistant collected the money, N&K
got what they wanted –everyone was
happy.
• I think this is probably true, but..
A Second Possibility
• N & K give too much detail, emphasise how
generous the elasticities used are, etc.
• Sometimes are very accurate, which they
emphasise, e.g. incomes in US cities,
sometimes very approximate, e.g. incomes in
German cities which are all the same, or London
which is attributed with the average UK income.
• Maybe all this detail is to persuade the reader
accurate calculations are being done, when they
aren’t, like the conjurer’s chatter.
Recalculation
• What would their figure look like if it had
been calculated the way they said it had?
• This has been done for the Australian and
European cities (except Moscow)
• The relationship of fuel use with density is
a lot less clear.
Other Urban Problems
• There are reasons for not recalculating:
• Moscow – in 1980 it was communist.
• Hong Kong – is a very small area and in
1980 one could not travel anywhere far by
car.
• Singapore is an island state with one road
route out to Johore.
Regression Analysis
From the ‘raw’ N&K data Ian Gordon shows:FU = 5.8 – 0.70d
(t= 9.5, Rsq = 0.78)
but with fuel price and incomes:FU = 6.7 – 0.23d - 0.75p + 0.25y
(3.7)
(6.0)
(0.7)
(Rsq =0.95)
The variables are natural logs so the coefficients
are elasticities (lower than N&K for p and y)
The Density Variable
• The regression does show density to be
important, but there are other problems.
• Firstly, there is N&K’s measurement of density.
For the US they use SMSAs which are larger in
area than the conurbation.
• For Europe they use political areas which are
smaller, e.g. the GLC excludes large areas of
outer London
Secondly, N&K disdain economic variables
• Their concern is with ‘parameters …in the
direct control of physical and transport
planners.’
• So they ignore the cost of public transport,
even when most European countries
subsidise.
• This means that the coefficient of density in
Gordon’s equation is very much an upper
estimate, the true figure is about half, i.e. about
0.12.
• This fits with an estimate by Peter Hall in 2001
that doubling density would reduce fuel use by
about 15%.
• Which, with Gordon’s estimated price elasticity,
would be achieved with a price increase of 20%!
Is higher density always good?
• Evidence from Norway and the
Netherlands suggests that high density
encourages long distance leisure trips.
• The minimum fuel usage is at 60 dwellings
to the hectare ( The English minimum was
30)
• But the evidence shows that the overall
savings from increasing density are not
great.
• (And it also shows that access to a garden
reduces leisure travel.)
British Planning Policy
• Build at high density on brown field sites
and contain settlements by the use of
Green Belts, etc.
• But this results in dispersion and high fuel
use.
Land
Value
Green Belts
Initial land value gradient AB
C
Urban expansion would lead to new gradient CD
A
Green Belt
Agricultural
Value
D
B
O
d1
d2
Green Belt prevents development between BD (d1 to d2)
Distance
Land
Value
New higher land value gradient EF, GH
E
C
A
Green Belt
F
Agricultural
G
D
B
O
Value
d1
H
d2
Distance
1.
Stated aims of green belts in planning terms:
i)
To check urban sprawl
ii)
To safeguard surrounding countryside
iii) To prevent towns merging
iv) To preserve the character of historic towns
v)
To assist in urban regeneration
Originally green belts were intended to be fairly narrow and to provide
recreation for the towns they surrounded (e.g. the Greater London
Plan of 1944)
There is now nothing about recreation and the belts now cover more
land than the towns they surround
2. In economic terms these planning aims should result in:
i)
Higher land and property values in the contained urban area
ii)
Commuting across the Green Belt
iii) Development on the other side of the Green Belt
iv) Amenity (and certainty) would lead to higher property values in
and on the edge of the Green Belt
v)
Infill and higher density development within the contained urban
area
(1) Brownfield Sites
• Does it make sense to build at high
density in the countryside because it is a
brownfield site?
• For example an old cottage hospital site
five miles west of Reading near Bradfield
where a bus stops 800m away about once
an hour.
(2) Containment round Oxford
• Oxfordshire planned that new housing
would be outside Oxford at smaller towns
with good public transport.
• The result, increased use of cars to get to
work.
• But new housing adjacent to Oxford did
not result in more car use.
New Homes in England
• Between 2000 and 2004 there were
500,000 dwellings added to the housing
stock in England
• Of which 300,000 were in urban areas
• And 200,000 in rural areas, a large
number being conversions and infill in
villages
US Evidence
• Where policies of containment are in force
fuel use is greater.
• As people commute from one contained
area to another.
Conclusion
• The policies that we have are justified by
reference to ‘global sustainability’.
• But the evidence for these policies as
reducing fuel use and emissions is nonexistent at best.
• And there is evidence that they worsen the
situation.
• We know that British policies are there for
other reasons – some people think high
densities are good per se, some people
just want to protect the countryside.
• But can we have a little honesty?