Trends in income and price elasticities of transport demand (1850–2010)
Roger Fouquet a,b,n
a
Basque Centre for Climate Change (BC3), Alameda Urquijo 4-4, 48008 Bilbao, Spain
b
IKERBASQUE, Basque Foundation for Science, Alameda Urquijo 36-5, 48011 Bilbao, Spain
a r t i c l e i n f o
Article history:
Received 5 December 2011
Accepted 1 March 2012
Available online 22 March 2012
Keywords:
Income and price elasticities of demand for
transport
Energy transitions
Rebound effects
a b s t r a c t
The purpose of this paper is to estimate trends in income and price elasticities and to offer insights for
the future growth in transport use, with particular emphasis on the impact of energy and technological
transitions. The results indicate that income and price elasticities of passenger transport demand in the
United Kingdom were very large (3.1 and À1.5, respectively) in the mid-nineteenth century, and
declined since then. In 2010, long run income and price elasticity of aggregate land transport demand
were estimated to be 0.8 and À0.6. These trends suggest that future elasticities related to transport
demand in developed economies may decline very gradually and, in developing economies, where
elasticities are often larger, they will probably decline more rapidly as the economies develop. Because
of the declining trends in elasticities, future energy and technological transitions are not likely to
generate the growth rates in energy consumption that occurred following transitions in the nineteenth
century. Nevertheless, energy and technological transitions, such as the car and the airplane, appear to
have delayed and probably will delay declining trends in income and price elasticity of aggregate
transport demand.
& 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Travel over the last two hundred years has been revolutionised
by a series of technological innovations. The shift from horsedrawn
transport to railways to cars has radically changed people’s
lives. Alongside these shifts in modes of transport were energy
transitions, from horse-power to coal to petroleum. Radical
changes in the technology and energy sources suggest upheavals
in transport behavioural patterns. This would imply difﬁculties in
understanding and anticipating the impact of any future technological
and energy transitions, such as a possible switch to electric
vehicles.
One particularly interesting observation is that energy transitions
are characterised by major increases in energy consumption
(Gr ¨ubler et al., 1999; Gr ¨ubler, 2004). If this was true for future
transitions, it would have major implications for long run demand
for energy and carbon dioxide emissions. A transition to, for
instance, electric vehicles might, therefore, produce major
increases in electricity demand (beyond the expected substitution
from petroleum), and may lead to large emissions if it was
generated from fossil fuels.
While past energy and technological transitions have been
disruptive, consumers before and after were still demanding
transport services. The same should apply to future transitions.
In other words, focussing on the long run aggregate demand for
transport services might offer a way to identify stable patterns of
behaviour in periods of transitions. In addition, estimating the
long run demand for transport, and how it has evolved, might
bring some light onto whether future energy transitions will be
similar to past ones.
Few studies have investigated the long run evolution in the
demand for transport or other energy services. Fouquet (2008 p.
265) proposed that the nineteenth century in the United Kingdom
was a period of both energy transitions and of very high income
and price elasticities, especially related to lighting and passenger
transportation. However, at the time, no econometric analysis
was performed to test these propositions.
Small and van Dender (2007) looked at transport demand in
the US between 1960 and 2004, and suggested that price
elasticity was lower between 2000 and 2004 than from 1960 to
2000. Goodwin et al. (2004) surveyed the literature on transport
demand elasticities and tentatively concluded that income elasticities
may have declined over the last forty years. Fouquet and
Pearson (2012) offered estimates of trends in the demand for
energy services, focussing on lighting, for a longer period
(between 1750 and 2008), which is probably more relevant for
studying energy transitions. This study found that income and
price elasticities increased dramatically (to 3.5 and À1.7,
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/enpol
Energy Policy
0301-4215/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enpol.2012.03.001
n
Correspondence address: Basque Centre for Climate Change (BC3), Alameda
Urquijo 4-4, 48008 Bilbao, Spain. Tel.: þ34 944014690; fax: þ34 944054787.
E-mail address: roger.fouquet@bc3research.org
Energy Policy 50 (2012) 62–71
respectively) between the 1840s and the 1890s and fell rapidly in
the twentieth century.
Using the same approach, this paper estimates the trends in
income and real price elasticities of demand for passenger
transport. An understanding of trends in transport demand
elasticities in the United Kingdom will also be valuable for
anticipating future transport use, energy demand and carbon
dioxide emissions in this economy and others with similar levels
of economic development and characteristics related to transport.
For instance, if price elasticities for transport demand are low and
falling, then high taxes on (or prices of) carbon are likely to have
little effect on overall transport use. However, they may signiﬁcantly
encourage the use of low carbon technologies, energy
sources and behaviour. Furthermore, this historical perspective,
looking at the United Kingdom at different phases of economic
development, may also be of relevance for understanding future
trends in transport demand in developing economies.
Section 2 brieﬂy reviews the literature on the demand for
passenger transport. Section 3 outlines, the data sources for this
study. Then, Section 4 presents and discusses the trends in
transport prices and consumption, as well as income. In Section
5, the estimated trends in the price and income elasticities of
demand for transport are presented. The ﬁnal section draws
tentative conclusions about the variations in elasticities over
time, and examines their implications for our understanding of
energy transitions, long run energy consumption and climate
policy.
2. Demand for transport services
The demand for passenger transport services reﬂects individuals’
willingness to pay for travelling from one place to another.
Before the nineteenth century, as well as having very limited
incomes, people had lifestyles that required little transport. Work
was in or near the home. People socialised with their family and
neighbours. Yet, historically and across cultures people’s travel
distances have been bound by their monetary and time budgets
(Sch¨afer, 2000). Greater income allowed individuals to spend
more money on travel. Similarly, new technologies and cheaper
travel created an opportunity for the location of work and
lifestyles to change (Gr ¨ubler et al., 1999; Bannister, 2011).
Although some evidence supports a constant relationship
between income and travel demand (Sch¨afer and Victor, 2000),
this crucial issue in future energy use and carbon dioxide emissions
deserves a deeper look.
Consumer responsiveness to changes in prices and income has
depended on a number of factors. Income elasticities tend to
reﬂect whether consumers perceived a particular good as a
‘‘luxury’’ or as a necessity. For so-called ‘‘luxury’’ goods and
services, consumption increased more than proportionally as
income rose (i.e., high income elasticity). Normal goods and
services (or necessities) were likely to have low elasticities (i.e.,
less than 1). In some cases, they were seen as inferior goods and
services, and consumption declined with greater income. The
traditional view is that, at low levels of economic development,
most goods and services have been luxuries relative to basic
foods. So, as incomes rose, except for basic foods, consumption for
all goods and services increased more than proportionally. As
income increased further, saturation effects implied that consumption
and expenditure of many previously ‘‘luxury’’ goods
and services grew less than income (Moneta and Chai, 2010).
Evidence suggests that expenditure categories mostly associated
with services have tended to experience relatively lower
levels of saturation. This is partly explained by the introduction of
higher quality services delaying the saturation effect. Consistent
with this, there is some, but limited, evidence of saturation for
travel services (Moneta and Chai, 2010). In other words, as an
economy develops and incomes rise, income elasticities associated
with travel service demand might be expected to fall, but
not necessarily to zero.
Price elasticity can be broken-down into the income and
substitution effects. The income effect depends on the proportion
of the individual’s budget spent on the service and whether the
individual tends to spend a greater proportion of the budget on
the service as income rises. The substitution effect indicates the
tendency to switch towards the cheaper good or service. The
more substitutes available for a particular service the greater will
be the substitution effect.
When considering the aggregated market for passenger transport
services, it is difﬁcult to identify clear substitutes. One could
argue that communication, such as postal services, the telegraph,
the telephone and emails, has offered a partial substitute, by
being able to transmit messages from one person to another
without needing people to travel. Thus, cheaper, faster and better
communication over the last two hundred years may have
reduced the need for certain types of travel (Selvanathan and
Selvanathan, 1994). At the same time, they (and especially other
forms of communication, such as newspapers, radio, television
and the internet) have probably increased people’s desire to travel
more generally (Salomon, 1985). Even today, with emails, there
still appears to be a relationship between distance and the mode
of communication used – for interactions with people living less
than 5 km, people still often engage in face-to-face communication;
for regional interactions, the phone is used mostly; and for
greater distances, the internet is the predominant choice (Mok
et al., 2010).
It is important to distinguish between the overall market
elasticity of demand for transport and the demand facing individual
modes of transport. Here, the focus is on the overall demand
to identify a continuous demand for travel over different modes,
technologies and energy sources. Most studies, however, focus on
a particular mode of transport, rather than aggregate demand. An
early survey of the literature offers an insight into some of the
elasticities that might be expected by modes of transport (Oum
et al., 1990, see also Oum et al., 1992). For rail travel, it identiﬁes a
wide range of price elasticities between À0.11 and À1.80. The
variation is explained in great part by the purpose and type of rail
travel. Average peak intra-city travel was À0.15 – that is, a 10%
rise (or fall) in prices only reduced (or increased) consumption
1.5%. Whereas urban off-peak rail price elasticity was À1. Intercity
business travel was À0.8, and average leisure rail travel was
estimated to be À1.4. For off-peak car travel, price elasticities
ranged from À0.06 to À0.88 and, for peak journeys, estimates
were between À0.12 and À0.49. Peak bus demand was inelastic
(i.e., 0), while off-peak demand ranged from À1.08 to À1.54. Out
of interest, the authors also found that the price elasticity of
demand for air travel ranged from À0.08 to À4.51, indicating, in
some cases, great sensitivity to prices for this more ‘‘luxury’’ form
of transport. However, this study was unable to distinguish
between short run and long run estimates. Also, these estimates
are for developed economies, while the current paper is interested
in demand at different phases of economic development.
More recently, Goodwin et al. (2004) surveyed the literature
associated with car travel. In addition to providing a valuable
range of estimates, they were interested in changes in elasticities.
Although their results depend on the assumptions made about a
traveller’s utility function, their theoretical expectations (presented
in Hanly et al., 2002) were that price elasticities increased
when (fuel) prices were high and fell with lower prices, and that
price elasticities fell with income (and, thus, over time). One
might also expect a decline in price elasticities because, for
R. Fouquet / Energy Policy 50 (2012) 62–71 63
certain journeys, transport and leisure are consumed jointly and
the travel costs becomes a declining share of the total costs.
However, they found no clear support for these expectations. On
the other hand, they proposed that, because of saturation effects,
income elasticities might be expected to fall, and tentatively
concluded that income elasticities did fall.
Price elasticity is also of great interest to energy economists
because of possible rebound effects (Howarth, 1997; Greening
et al., 2000; Sorrell, 2007). Jevons (1865) introduced the concept
of the rebound effect, arguing that improvements in energy
efﬁciency were likely to lead to greater energy consumption
(not less). Ayres (2005) proposed that rebound effects for ‘‘macro’’
innovations (i.e., radical innovations, like the steam engine) might
generate large rebound effects and increases in energy consumption;
and ‘micro’ innovations that improve the efﬁciency of
existing technologies result in smaller rebound effects. Fouquet
(2008, p. 277) proposed a few historical cases in the United
Kingdom (particularly, freight transport between 1715 and the
1930s, passenger transport from the 1840s to the 1920s, and
lighting during the nineteenth century – the latter conﬁrmed in
Fouquet and Pearson (2012)) where the rebound effects were very
high – partially supporting Jevons’ (1865) hypothesis.
In this light, Small and van Dender (2007) examined the price
elasticity of demand for car transport. They looked at transport
demand in the US between 1960 and 2004, and found that price
elasticities were lower between 2000 and 2004 than from 1960 to
2000. They suggest that the rebound effect fell from 2.1% to 0.57%
for a 10% efﬁciency improvement. Thus, for this limited sample,
efﬁciency improvements generated only minor increases in car
travel and considerable reductions in energy use. Hughes et al.
(2006) also ﬁnd that the price elasticity of gasoline demand has
declined through time, but that, at any particular time, higher
incomes groups have higher price elasticities than lower income
groups. The very limited evidence on trends in income and price
elasticity of demand for transport demand, and related rebound
effects, invites a more detailed time series study. The rest of this
paper seeks to provide more evidence.
3. Data sources and creation
To study the relationship between passenger transport use,
income and prices, it is necessary to gather statistical information
on travel and the prices of travelling or energy sources associated
with the different modes. This section offers a summary of the
sources and methods used to produce annual estimates of prices
(or costs) and use of horse drawn transport (1850–1924), railways
(1850–2010), buses (1904–2010) and cars (1904–2010) in Great
Britain and then, from the 1920s, the United Kingdom – more
detail can be found in Fouquet (2008).
For horse-drawn transport, Chartres and Turnbull (1983, p. 71)
presented estimates of passenger miles per week for a number of
years between 1715 and 1840. Thompson (1976) offered detailed
estimates of the number of horses associated with different
activities (stagecoaches, carriage, riding, as well as farm and
trade), during the nineteenth and early twentieth century, and
particularly from 1851. These can be linked to the 1840 estimate
by Chartres and Turnbull (1983, p. 71) to estimate the average
number of passenger kilometres per horse used. With this
average, the number of horses used for transport can be used
for calculating the millions or billions of passenger-kilometres
(bpk) from 1851 to 1924. While this was a rough estimate, in the
second half of the nineteenth century, horse transport was
dwarfed by railways use.
For prices, Jackman (1960) collected considerable information
on the cost of stage coach travel in the eighteenth and nineteenth
centuries. When these were divided by the distance for particular
stage coach journeys, they produced estimates of the price per
passenger-kilometre. From estimates of several of the main
journeys in England, an average price was estimated.
For railway use, Hawke (1970, p. 47) offered data on millions
of passenger miles and millions of journeys travelled between
1840 and 1870. Mitchell (1988, p. 545) also presented data on
millions of journeys travelled annually from 1842 until 1913.
Thus, to calculate passenger-kilometres after 1870, it was necessary
to have estimates of the distance travelled during an average
journey. Munby (1978, pp. 106–7) gave estimates of billions of
passenger-kilometres and the average length of train journeys
between 1920 and 1970. So, the gap was only between 1870 and
1920. Surprisingly, the calculation of the average length of a
railway journey in 1870 (from Hawke, 1970) and in 1920 (from
Munby, 1978) are virtually identical, and the distance travelled on
the average railway journey was assumed to have remained
nearly constant over those ﬁfty years. Mitchell’s (1988) passenger
journey data was multiplied by the average length of journeys to
estimate passenger transport in billions of passenger-kilometres
(bpk) between 1870 and 1913. To complement Munby (1978),
from 1938 to 2010, DoT (2002) and DfT (2011) presented direct
estimates of passenger-km.
These can then also be used to estimate the price of railway
passenger services. Mitchell (1988, p. 545) has railway passenger
receipts (in £m) between 1843 and 1980. These ﬁgures can be
divided by estimates of total passenger-km travelled (discussed
above) to suggest the cost per passenger-km of using railways up
to 1913. Munby (1978, pp. 113–114) has direct estimates of the
price (in old pence per km) from 1920 to 1970. Between 1970 and
1980, receipts were again divided by passenger-km. DoT (2002)
and DfT (2011) had estimates of the price between 1991 and
2010. An inability to ﬁnd values during the 1980s led to
interpolation.
For twentieth century road travel, DoT (2002) and DfT (2010)
presented data on passenger travel (e.g., buses, cars, motorcycles,
etc.) between 1952 and 2010. Before the early 1950s, data needed
to be pulled together. Mitchell (1988, pp. 557–558) had information
on the number of cars and other vehicles between 1904 and
1980. Car passenger travel per vehicle was calculated for 1952 –
16,000 km per car per year. Before these years, assumptions were
made about the trend in average travel per vehicle: for cars, an
annual 2% decrease in the average travel per car ratio each year
before 1952 – based on the idea that there were fewer roads and
people travelled less per year than they did in the 1950s. This
assumed trend in average distance per vehicle (back to the
beginning of the century) was multiplied by Mitchell’s (1988,
pp. 557–558) numbers of vehicles to produce an estimate of the
bpk back to 1904. Given that the number of buses was also
available, a similar method was used to quantify bpk used related
to public transport vehicles (Fouquet, 2008).
Passenger road transport ‘‘prices’’ are more complex in the
second-half of the twentieth century. Rather than identifying the
cost of buying a train or bus ticket, as more travellers used their
own cars, a number of different expenditures were involved. At
least three costs can be identiﬁed: the fuel costs, all marginal
costs and annualised total costs. Table 1 shows the breakdown of
the annual cost of car travel between 1971 and 2008. Fuel costs
accounted for between 28% and 40% of the total annual expenses.
Few of the other expenses listed are obvious marginal costs – tyre
consumption also depends on distance travelled. So, fuel costs
were presented as the price (or main private marginal cost) of
passenger transport.
The price of driving a car one kilometre was estimated by
dividing passenger fuel expenditure (in million tonnes of oil
equivalent (mtoe)) by distance travelled (in bpk). Until the
R. Fouquet / Energy Policy 50 (2012) 62–7164
1930s, motor spirit (i.e., gasoline) was the main petroleum
product for road vehicles. Then, diesel (that is, ‘‘derv’’) began to
be used by goods vehicles and buses. Cars used exclusively motor
spirit until the 1980s, when diesel started to take a small market
share. Thus, it was necessary to identify the share of motor spirit
and diesel consumed for passenger services.
The Ministry of Power (MoP, 1961) indicated the share of
motor spirit used by both passenger and freight transport
between 1938 and 1960. The share of motor spirit used by
commercial vehicles fell from 42% to 19% in that period – it was
assumed that the share of buses was around 10% of commercial
vehicles in 1938 and 5% by 1960, reﬂecting the relative decline of
public transport. These shares enabled motor spirit consumption
(DTI, 1997,2001,2007,2011 and back copies; MoP, 1961 and King,
1952 p. 551) to be estimated for passenger travel.
DTI (1997,2001,2007) presented the amount of diesel used by
passenger travel back to 1995; in 2000, it was equivalent to 18%
of the total consumption of diesel. Miller (1993) indicated the
share of diesel vehicles in the passenger vehicle stock into the
early 1990s, providing a basis for calculating the small amount of
diesel consumed. An interpolation connected 1991 and 1995. The
rest of road transport diesel (also found in DTI, 2007; MoP, 1961
and King, 1952, p. 551) was assumed to be for freight. Thus,
consumption of motor spirit and derv for passenger road transport
were estimated between 1910 and 2000.
DTI (2011, 1997, 2001) presented the price of motor spirit and
diesel back to 1954. The Institute of Petroleum (1994) had data on
the price of motor spirit from 1902 to 1953. The prices were
multiplied to the consumption estimates to calculate the fuel
expenditure of passenger road transport. Dividing fuel expenditure
by distance enabled an estimate of fuel costs of road
transport services in pence per passenger-km. All prices for
transport service were converted into real terms using the data
available in Allen (2007).
4. Trends in transport prices and use
Passenger transport in Britain experienced a series of revolutions
that radically altered the ability to travel. First, from the
mid-seventeenth century, the introduction of turnpikes, managed
roads that travellers paid to use, enabled an improved road
network. Then, from the 1770s, stage coach journeys became
better managed and faster – by 1830, average trips took one-ﬁfth
of the time they took in the 1770s (Bagwell, 1974). Between 1775
and 1815, passenger travel increased from 0.05 billion passengerkm
(bpk) to 3 bpk – a 60-fold increase in 50 years (Chartres and
Turnbull, 1983).
Then, harnessing the power of steam by heating fuels revolutionised
the economy and society. It, ﬁrst, provided stationary
power to remove water from coal mines and to spin cotton, then,
as directed power, enabled transport along railways (and for
ships). The introduction of steam engines to rail in the ﬁrst half
of the nineteenth century meant that coal rather than horses
could power the carriage of people and goods. In Britain, coal was
cheap (see Fig. 1) and entrepreneurs were willing to build
railways that would eventually form a vast network – increasing
16-fold in the 1830s, and then another four times in the 1840s, it
expanded from around 150 km in 1830 to nearly 10,000 km in
1850 and then to 30,000 km by 1900 (Mitchell, 1988, p. 541).
For the wealthy and middle classes, the railway was a superior
way to travel medium and longer distances. In addition to the
relative improvement in comfort, per passenger-km, in the 1830s
and early 1840s, stage-coach travel cost more than 70 pence
(2000 prices) and the railways around 30 pence (Fouquet, 2008 p.
163). It appears, however, that coach companies, fearing competition
from railways and having invested in equipment and horses,
drove up their prices in the 1820s (Hart, 1960). Inevitably, this
strategy accelerated the transition to railways. This meant that
the average price of transport fell greatly during the 1830s and
1840s (see Fig. 2). Railway use increased nearly ﬁvefold in the
1840s to 1.5 bpk (see Fig. 3). For shorter distances, including to
the railway station, horses were still needed. So, transport more
generally increased to an estimated 5 bpk in 1850.
Table 1
Shares of motoring costs for average households, 1971–2008.
Source: based on data from DfT (2010) and ONS (2002).
1971
(%)
1981
(%)
1991
(%)
1996
(%)
2001
(%)
2006
(%)
2008
(%)
Purchase, tax &
insurance
53.3 55.0 60.1 53.3 52.2 44.9 39.9
Spares, repairs &
servicing
14.5 15.3 12.5 14.1 14.2 15.4 16.3
Petrol, diesel &
engine oils
27.9 33.3 24.7 28.5 30.0 34.9 39.7
Other motoring
costs
4.3 2.7 2.5 4.1 3.6 4.6 4.0
Motor Spirit/
Gasoline
Coal
Provender
(for Horses)
Fig. 1. Trends in the prices of energy sources for transport, 1800–2010.
Source: Fouquet (2011).
0
10
20
30
40
50
60
70
80
90
100
0
500
1,000
1,500
2,000
2,500
3,000
3,500
1300
pence(2000)perpassenger-km
£(2000)pertonneofoilequivalent
Energy for
Transport
Transport
* Five-year averages
1400 1500 1600 1700 1800 1900 2000
Fig. 2. Trends in the prices of energy for transport and of transport*, 1300–2000.
Source: Fouquet (2011).
R. Fouquet / Energy Policy 50 (2012) 62–71 65
A number of factors were driving the consumption in transport
services in the second half of the nineteenth century. Income
levels increased substantially – more than doubling in the
second-half of the nineteenth century (see Fig. 4). More efﬁcient
steam engines in the 1860s and 1870s helped. In 1840, one tonne
of coal could achieve 1700 passenger-km; by 1870, it could move
more than 3000 passenger-km; and 3400 passenger-km in 1900
(Fouquet, 2008 pp.165–166). Although important, fuel costs were
a relatively small share of the price of transport – falling from
over four pence to under two pence per passenger-km, which
amounted to about one-tenth of the price (see Fig. 2). Better
management enabled travel prices to fall. For instance, by 1842,
the railway network had become commercially integrated, allowing
customers to buy one ticket for connecting journeys. Also,
speed of travel increased and comfort improved (Leunig, 2006).
Furthermore, in the latter part of the nineteenth century,
owners began to appreciate, with governmental pressure, that
third-class customers were a large potential market. Selling cheap
tickets allowed a large proportion of the population, whose only
other option was walking (or possibly riding on a slow wagon), to
travel more and save time, opening-up many opportunities
(Leunig, 2006, p. 651). The railways, which started as an improvement
on the stagecoach for the wealthy, became a new service for
the masses, transforming the nature of mobility and society.
Consequently, passenger transport increased from 5 bpk in 1850
to nearly 30 bpk in 1900 – with railways providing 63% of the
services (see Fig. 3).
Interestingly, since the distance travelled during the average
train journey appears to have remained constant between 1870
and 1920 (see Section 3), the growth may have been associated
only with an increase in the number of journeys travelled.
Alternatively, there may have been far more shorter (e.g., intraurban)
and longer (e.g., inter-city) journeys, and they happened to
have cancelled each other out, leaving the average distance of a
journey unchanged.
The next transition was the result of the introduction and
development of the internal combustion engine. Despite the
greater cost per unit of energy (see Fig. 2), petroleum derivatives,
such as gasoline or motor spirit and diesel or derv, became the
dominant energy sources for the internal combustion engine, and
eventually transportation.
For the middle and lower classes, the bus offered a cheap and
more ﬂexible public transport service than horses for shorter
journeys, and improved connections between train stations
(Bagwell, 1974, p. 224). Despite competition from the incumbent
tramways in many of the larger cities, their use increased
spectacularly in the ﬁrst-quarter of the twentieth century.
Between 1903 and 1913, there was a 270-fold increase in
motor-buses used in London (Munby, 1978, p. 562). By 1930,
there were more than 50,000 buses on British roads, providing an
estimated 50 bpk. This increased in 1950 to just under 80,000
buses, providing nearly 90 bpk by 1950 (Mitchell, 1988, p. 557–
558; DfT, 2010).
Meanwhile, the car started as a toy for the rich. The number of
private motor cars owned in Britain grew rapidly from 8000 in
1904 (the ﬁrst year for which data exists) to 132,000 in 1914 – a
16-fold rise in ten years (Mitchell, 1988, p. 557). Then, after the
First World War, and with the rising per capita income levels (see
Fig. 4) and falling car prices (the average price of a car fell from
about £(2000) 25,000 in 1910 to £(2000)10,000 by the early
1930s (Foreman-Peck, 1981, pp. 264–265)), middle class car use
grew. By 1924, 0.5 million cars were being driven and, in 1938,
2 million cars (Mitchell, 1988). Car travel increased from 0.55 bpk
in 1910 to 2.5 bpk in 1920 to 15 bpk in 1930 to 32 bpk in 1939 –
dropping substantially during the Second World War (see Fig. 3).
Interestingly, between 1924 and 1938, motor spirit consumption
increased from 1.6 to 4.4 million tonnes. This implied that, by
the end of this period, each car consumed 2300 l per year, which
was a 50% decline on use 14 years earlier (BPP, 1926; King, 1952).
This indicated that, in the 1920s and 1930s, fuel efﬁciency and
average travel habits changed greatly.
Several factors were driving-up the demand for and use of car
travel. From the 1920s until the early 1970s, motor spirit prices
remained relatively stable, averaging about £(2000) 0.65 per litre.
Despite the Oil Shocks, average oil prices stayed within the £(2000)
0.50–0.80 per litre range (or £(2000) 700–11,000 per tonne shown
in Fig. 1) in the second half of the twentieth century (Fouquet, 2008,
p. 177). Second, fuel efﬁciency (measured as the passenger-km per
tonne of oil used) improved 86% in the third quarter of the twentieth
century. By 1990, fuel efﬁciency improved another 15%. However, it
fell 4% over the next decade – most probably reﬂecting the fashion
for bigger and more powerful vehicles. Finally, per capita income
increased 76% between 1950 and 1975, and 75% between 1975 and
2000 (see Fig. 4).
Despite rising incomes, bus use fell 32% from 1950 to 1975,
and 25% from 1975 to 2000. Similarly, passenger rail travel use
remained stable in the third-quarter of the twentieth century, and
then rose 30% in the ﬁnal quarter. In the meantime, car ownership
soared from 2.2 million in 1950 to 14 million in 1975 (Mitchell,
1988, p. 557–558) to 23 million in 2000. Car travel rose at a
similar rate – from 50 bpk in 1950 to 330 bpk in 1975 (i.e., nearly
Horse-Drawn
Railways
Bus
Car
Airplane
Fig. 3. Consumption of passenger transport, 1750–2010.
GDP per Capita
Population
Fig. 4. GDP per capita and population, 1700–2000.
Source: ONS (2002), Mitchell (1988), Broadberry et al. (2009).
R. Fouquet / Energy Policy 50 (2012) 62–7166
a sevenfold increase) to 640 bpk in 2000 (i.e., nearly a doubling)
(see Fig. 3).
This suggests that travel behaviour changed considerably
during this time. In addition to declining motoring costs and
rising income, the nature of transport was radically transformed
by the introduction and expansion of cars. For instance, cars
allowed freedom to determine the destination, ﬂexibility about
the route taken and privacy (and exclusivity) during the journey.
The car began as a source of status and leisure, and still in the
1930s, three in every four cars bought were for leisure (O’Connell,
1998, p. 77). The growing use of cars created a cultural realignment
in which people expected to (and were expected to)
travel by car. Cities and related activities (whether associated
with work, shopping or leisure) increasingly became built around
accessibility by car (O’Connell, 1998). Thus, harder to measure
social and cultural forces also propelled the demand for car use.
An average car in 1950 was responsible for 15,000 passengerkm.
This increased 35% to 21,000 passenger-km in 1975 and then
another 30% to almost 28,000 passenger-km in 2000. Thus, in the
second-half of the twentieth century, mobility increased considerably
and all the growth in passenger transport was in the use of
private cars rather than buses or railways.
To summarise the long run trends in land transport, the
estimated average price of passenger transport services fell from
£(2000)0.63 per passenger-km in 1850 to £(2000)0.39 in 1900 to
£(2000)0.08 in 1950 to £(2000)0.05 in 2000 – a 12-fold decline in
150 years. Income per capita increased from £(2000)1500 per
year in 1850 to £(2000)3200 in 1900 to £(2000)5200 in 1950 to
£(2000)17,000 in 2000 – nearly a 12-fold rise in 150 years. Land
passenger transport consumption soared from less than 4.5 bpk in
1850 to 27 bpk in 1900 to 186 bpk in 1950 to nearly 740 bpk in
2000 – a 165-fold rise in 150 years. If income and price were both
unit elastic in relation to transport demand for these 150 years,
consumption would have increased 144-fold. Thus, the evidence
suggests the demand for land transport was either elastic for the
last 150 years or very elastic during certain periods.
Yet, the introduction of air travel has increased total passenger
transport even more rapidly over the last 60 years. In 1950,
airplanes provided 2 bpk. By 1975, this had increased to 62 bpk,
and up to 260 bpk in 2000, reaching 305 bpk in 2010. So, at
present, air travel provides almost half the number of passenger
kilometres that cars do – airplanes are responsible for 28% of all
passenger transport and cars for 61%. Recent total1
(i.e., land and
air) use increased more – from 192 bpk in 1950 to 1012 bpk in
2000, and 1095 bpk in 2010. Thus, total passenger transport
increased 220-fold in 150 years, indicating even higher income
and/or price elasticities.
5. Income and price elasticities, and rebound effects
This section presents some estimates of the inﬂuence of
income and prices on transport use, and their trends over the
last 150 years. Following the same approach as Fouquet and
Pearson (2012), a vector error correcting model was used to
provide an econometric analysis of the data and the trends, and
estimate the cointegrated relationship between travel, income
and transport prices. The emphasis should not be on the methods
used, which are open to criticism and other scholars might
improve upon,2
but the trends.
Given the trended nature of the data (see Figs. 2–4) and the
tendency for long run transport use (or energy related to transport),
GDP and travel costs to be cointegrated (Bentzen, 1994;
Fouquet et al., 1997; Ramanathan, 1999), the possibility of using
vector error-correcting models (VECM) was explored. From a
statistical perspective, such models were appropriate.
First, for the long run trends in transport consumption, GDP
per capita and the price of transport, non-stationarity could not
be rejected. In addition to the standard tests for unit roots, an
augmented Dickey–Fuller test where the time series is transformed
via a generalised least squares (GLS) regression was used
to improve the power of the test (Elliott, Rothenberg and Stock,
1996). Here, for up to 15 lags, and incorporating the assumption
of a time trend, the tau-statistics could not reject at the 10%
conﬁdence level. Thus, unit roots (i.e., non-stationarity) were
likely.
Second, the causal relationship between transport consumption
and per capita GDP was examined. The results suggest
unidirectional causality from per capita GDP to transport use. As
expected, GDP per capita was not inﬂuenced by either transport
use or prices, as transport has been only a small component of
economic activity (generally less than 8% of GDP). Although still
signiﬁcant, as will be shown by certain elasticity estimates, the
causality test indicated that prices were less of a powerful
explanatory variable of transport consumption. The results indicate
that, in the second-half of the nineteenth century, consumption
may have also inﬂuenced prices – perhaps reﬂecting the
importance of economies of scale in driving down early railway
costs. While this raised a simultaneity problem, for this exercise,
it was assumed that causality was unidirectional from prices to
consumption for the whole period.
Third, tests rejected the null hypothesis of no cointegrating
equations for the relationship between transport consumption,
GDP per capita and the price of transport – for the whole period
between 1850 and 2010 (see below). Having selected the appropriate
number of lags from a series of different tests (Nielsen,
2001), tests for the existence of cointegrating equations were
performed and, when the null hypothesis of no relationship was
rejected, almost always one cointegrated relationship could not
be rejected (based on methods developed in Johansen, 1988,
1995).
These VECM were, therefore, used to estimate the evolution of
income and price elasticities. The approach was to estimate
elasticities for ﬁfty year periods moving through time. For
example, the income and price elasticities were estimated for
the period 1850–1899, then 1851–1900, and so on until 1961–
2010. Then, the elasticity for any particular year would be the
moving average (i.e., the average of all elasticities estimated
where that year was included). For example, for the moving
average around the year 1950, ﬁfty income elasticity estimates
were produced (for the periods 1900–1950, 1901–1951, and so on
until 1950–2000) and the average of all these estimates was equal
to 0.91. Estimated in the same way, the average price elasticity
around 1950 was À0.72.
Inevitably, for some periods, the results were either not as
expected or the possibility of no cointegrated relationship could
not be rejected. In particular, the period broadly between 1850
and 1869 produced unexpected price elasticities and the absence
1
Sea passenger transport was excluded because the author has only managed
to ﬁnd data from 1950, despite its historical role in international transport. In
1950, there were 6.4 bpk by sea, implying that it provided less than 4% of total
passenger travel at the time. Its role fell further, providing less than 3 bpk at the
beginning of the twenty-ﬁrst century (DfT, 2011).
2
The author would like to thank Bill Nordhaus, David Stern, Lester Hunt and
Lutz Killian for their comments on this avenue of research. Lutz Killian questioned
whether long run elasticities can be estimated with time series, rather than with
panel data (see Kilian and Murphy, 2010). The other three scholars encouraged
this line of research, although offered caution about the methods. Naturally, the
author of this paper is solely responsible for the choices made and the limitations
of the methods used.
R. Fouquet / Energy Policy 50 (2012) 62–71 67
of cointegrating relationships could not be rejected. Most of the
price elasticities were positive and large. In the absence of a
satisfactory explanation, these estimates were excluded from the
calculation of the average estimates. Consequently, for the years
1850–1869 less than ﬁve estimates for price elasticity were
available and no average was calculated. Nevertheless, the majority
of estimates produced standard signs and sizes – income
elasticity estimates were used for 97% of the moving averages
over the 160 years and, between 1870 and 2010, price elasticity
estimates for 93% of the moving averages.
Fig. 5 presents the trends. It shows the clear the decline (in
absolute terms) of both income and price elasticities. In the third
quarter of the nineteenth century, they were at their highest,
falling in several waves, the ﬁrst, in the ﬁnal quarter of the
century, then, in the early 1900s, and, ﬁnally, from the 1920s until
the end of the century, a gradual decline for land transport.
It is worth noting ﬁrst, however, that consumption associated
with horse-drawn carriage in the late eighteenth and early nineteenth
centuries had increased 60-fold in ﬁfty years (Chartres and
Turnbull, 1983). In the last quarter of the eighteenth century,
average income grew 30%, travel prices fell around 25% and
journey speeds increased up to 80%. Thus, although there is a
lack of annual data, it indicates spectacular sensitivity to changes
in availability, prices, quality of service and income levels. So, we
can conclude that income (and other variable) elasticities were
very high, especially in the late 1700s.
In the 1850s, income elasticity was also very high – a 10%
increase in income led to a 31% increase in travel (see Table 2).
The creation and expansion of the railway network allowed the
upper and growing middle classes to travel more. This expansion
coincided with the rise in average income levels during the midnineteenth
century. Transport services were a key household
expenditure. During the second-half of the nineteenth century,
expenditure as a proportion of GDP increased from 6% to 7%
(Fouquet, 2008, p. 271). At the time, households were spending
three times more on travelling than on heating their homes and
six times more than on lighting their homes.
The growth in average income throughout the nineteenth century,
as well as the decline in prices, made railways accessible not only to
upper and middle classes, but also working classes. Fouquet and
Pearson (2012), commenting about lighting demand, proposed that
below certain levels of accessibility, the energy service only provides a
basic level of service. However, beyond a threshold, the service can
help to meet many other demands – in relation to work and to
leisure, thus, transforming lives.
For instance, the expansion of the railways coincided with
increased urbanisation. As populations moved to urban centres
from the beginning of the nineteenth century, they were closer to
transport nodes, thus, more able to take advantage of transport
services. Later, in the second-half of the nineteenth century, the
expansion of suburbia was driven by the growth in suburban
railway services. From the 1840s, upper and upper-middle class
households had sought to move away from the crime, sewage and
smoke, and the expansion of urban railway networks enabled
them to move to the suburbs (Luckin, 2000). The introduction of
the Cheap Trains Act of 1883 and a rapid expansion of suburban
housing in the 1890s offered an opportunity for lower-middle
class families to live in the suburbs and commute into the city
(Jackson, 2003; Burnett, 1986). The urbanised area of London
increased ﬁvefold between 1841 and 1901 (Demographia, 2011).
Thus, transport changed people’s lifestyles, which, in turn,
required higher levels of transport services to be sustained.
While transport services were still a ‘‘luxury’’, by the early
1900s, income elasticity of transport demand fell dramatically. It
fell from an average of 2.2 in the 1890s to 1.2 in the 1920s (see
Table 2). Surprisingly, the development of the internal combustion
engine and the growth of bus and car use did not stop the
decline in income elasticity. After the Second World War, land
transport became a ‘‘necessity’’, as it was an essential part of
everyday life.
It is surprising that the introduction of cars, allowing for
personal freedom of travel and privacy, did not raise income
elasticity. Nevertheless, without the introduction of cars, income
elasticity of travel demand would probably have fallen further,
implying a substantially less mobile economy and society. In
other words, the technological and energy transition associated
with the internal combustion engine and particularly the car,
which offered a different travelling experience, probably slowed
the decline in income elasticities.
This hypothesis (that energy and technological transitions can
delay the decline in income elasticities) is supported for air travel.
When the same regressions were run using total (land and air)
passenger transport, income elasticity stayed around 1.2 between
the 1920s and 1980s (see Table 2 and Fig. 5). NeverthePrice
Elasticities
Income Elasticities
Land and Air Transport
Land Transport
Fig. 5. Income and price elasticities for passenger transport demand, 1850–2010.
Source: see text.
Table 2
Decade averages of income and price elasticity of land and total passenger transport demand, 1850–2010.
Source: see text.
Decade Income elasticity Price elasticity Decade Income elasticity Price elasticity Decade Income elasticity Price elasticity
1850 3.1 – 1910 1.3 À1.0 1970 0.8, 1.2* À0.7
1860 2.9 À1.5 1920 1.2 À0.9 1980 0.8, 1.2* À0.7
1870 2.5 À1.5 1930 1.1, 1.2*
À0.8 1990 0.7, 1.1*
À0.7
1880 2.2 À1.3 1940 0.9, 1.0*
À0.8 2000 0.7, 1.0*
À0.6
1890 2.2 À1.1 1950 0.9, 1.2*
À0.7 2010 0.8, 1.0*
À0.6
1900 1.8 À1.1 1960 0.9, 1.2*
À0.7
Note (*
): ﬁrst value is for land transport; second value is for total (land and air) transport.
R. Fouquet / Energy Policy 50 (2012) 62–7168
less, looking in a little more details indicates that income
elasticities for total passenger transport started to decline from
the mid-1960s, but only reached unity at the beginning of the
twenty-ﬁrst century. For land passenger transport, an income
elasticity of 1 was reached in the 1940s. Thus, it appears that the
introduction of air travel delayed the decline in passenger transport
income elasticity by roughly 60 years.
Interestingly, air travel had far less inﬂuence over price
elasticities. It very slightly reduced consumer sensitivity to prices.
This may be because consumers are more driven by their rising
incomes. More probably, however, it simply reﬂects the price
variable used. This was an average of land passenger transport,
and does not incorporate the declining price of air travel between
1950 and 2010, for which data was not found.
Returning to the long run trends, the general decline in land
transport price elasticities was more gradual than for income (see
Table 2). In the 1870s, price elasticity was À1.5, implying that a
10% decrease in transport prices would lead to a 15% rise in
passenger travel (during the 1850s and 1860s, only a few of the
price elasticity estimates were signiﬁcant). This fell close to À1
by 1890, falling to À0.9 during the 1920s and to À0.7 for most of
the second-half of the twentieth century. Between 2000 and 2010,
price elasticity stood at À0.6. This indicates that, over the last 60
years (and throughout the major shift to cars, the dependence on
oil for transport and the Oil Shocks), a surprisingly stable relationship
has existed between transport (and indirectly petrol) prices
and consumption.
Separating price elasticity into its two components can help a
little to understand the decline. The income effects appear to have
fallen – travel expenditure relative to GDP fell from more than 8%
in the 1920s down to 2% at the end of the twentieth century
(Fouquet, 2008, p. 271). So, changes in prices would have had a
much greater effect on consumer purchasing power at the
beginning of the twentieth century than at the beginning of the
twenty-ﬁrst century. Given the difﬁculty of identifying clear
substitutes for transportation, it is hard to assess whether the
substitution effect changed over the last 150 years. It is possible
that communication technologies could have acted as a substitute
for certain transport needs. However, the telephone, for instance,
was very slow to diffuse in the United Kingdom – although
introduced in 1881, only 21% of households had telephones by
1965; this increased to 77% in 1983 (Bowden and Offer, 1994, p.
744). One might propose that, during the twentieth century,
while income effects were falling, the substitution potential was
increasing. Thus, the considerable changes in income effects and
possibly substitution effects over the last 80 years may have
cancelled each other out (to a certain extent), implying that there
was only a modest decline in price elasticities.
Transport demand price elasticities are also of interest for
identifying possible rebound effects. Energy efﬁciency improvements
associated with cars effectively reduced the marginal cost
of travel. Although this was not very relevant for the nineteenth
century, since the fuel costs were a small proportion of the price
of rail transport, it has been more important in the second-half of
the twentieth century or the twenty-ﬁrst century. The study
suggests that, despite the Oil Shocks and major changes in car
technology, consumers today may still be increasing their travel
by 6% for a 10% increase in fuel efﬁciency, implying only a 4%
energy saving. That is, this study proposes that there is a
substantial rebound effect associated with aggregate transport
demand.
This is considerably larger than other studies, such as Small and
van Dender (2007). One explanation is that, in fact, the rebound
effect will be smaller than 6%, because fuel costs (if one of the main
marginal costs) tend to be less than one-third (although 40% in
2008) of the total annual expenditure on cars (see Table 1). So, the
rebound effect may be considerably lower – that is, closer to 2%,
implying a 8% energy saving. Another possible explanation is that
studies that focus only on car travel will tend to ignore the effects
between modes of transport – such as cheaper costs of running a car
encourage rail users to travel more by car.
6. Conclusion
This paper sought to estimate the trends in real income and
price elasticities of demand for aggregate transport. Focussing on
the experience in the United Kngdom, it tried to identify the
inﬂuences of the increase in per capita income (12-fold between
1850 and 2000) and of the decline in the real price of transport
(12-fold between 1850 and 2000) on the rapid rise in aggregate
passenger land transport consumption (165-fold over 150 years).
Using standard econometric techniques and modelling of long run
economic behaviour, a series of income and price elasticities were
estimated. For each year, a moving average estimate was calculated,
thus, providing a time series of the income and price
elasticities from the middle of the nineteenth century to the
beginning of the twenty-ﬁrst century.
The results indicate that income and price elasticities were
very large (3.1 and À1.5, respectively) in the mid-nineteenth
century and declined continuously since then. Land transport
demand became price inelastic (i.e., less than one) in the early
1920s and income inelastic at the end of the 1930s and in the
early 1940s. However, the trends in income elasticity for total
passenger transport demand remained higher, around 1.2, and
only fell to unity in the twenty-ﬁrst century. This indicated that
the introduction of air travel delayed the decline in income
elasticities by sixty years.
It is worth stressing points made earlier about the study’s
limitations. The reliability and coverage of the data (especially for
the nineteenth century) and also the validity of the econometric
procedures can be questioned and criticised. Perhaps the estimates
are the outcome of a ﬂawed statistical analysis. Yet, the
econometric estimates broadly match the data – that is, between
1850 and 1900, jointly income and price elasticities were greater
than one and, between 1950 and 2000, for land transport, they
were less than one.
Also, providing a long-run perspective on changes in transport
demand elastiticies, these results support theoretical and prior
empirical expectations about transport demand. These expectations
were, ﬁrst, that saturation effects will reduce income
elasticity, although more slowly than for basic goods (Hanly
et al., 2002; Moneta and Chai, 2010; Goodwin et al., 2004) and,
second, that price elasticities fall as prices decline and incomes
rise (Hanly et al., 2002; Small and van Dender, 2007). They are
also similar to the trends in elasticities for lighting demand –
income and price elasticities peaked in the second-half of the
nineteenth century (3.5 and À1.7, respectively) and declined
during the twentieth century (Fouquet and Pearson, 2012).
As well as matching expectations and providing greater detail
on trends, these results are of interest for anticipating future
behaviour. The elasticity estimates, which were remarkably stable
during the twentieth century, suggest that aggregate passenger
land transport consumption in the United Kingdom will increase
by 8% from a 10% rise in income and by 6% from a 10% decline in
average transport prices in the early twenty-ﬁrst century. This
also gives clues to the rebound effect, which is estimated to be
between 2% and 6% from a 10% fuel efﬁciency improvement.
Since the study offers a long run perspective, some readers
may be interested in long run forecasts of trends in elasticities. It
is obvious that to do so is to risk being very wrong. However,
current forecasts of energy demand and carbon emissions look
R. Fouquet / Energy Policy 50 (2012) 62–71 69
forward decades and centuries, and depend on some evidence
about the trends in elasticities. So, despite the speculative nature
of forecasting (and the slight increase in income elasticities
during the ﬁrst decade of the twenty-ﬁrst century), the trends
in this study and prior expectations suggest that income and price
elasticities are likely to decline gradually.
One of the reasons for focusing on aggregate transport demand
and the related elasticity trends was to identify the impact of energy
transitions on energy use. Aggregate transport demand offered a
stable variable through periods of dramatic change. It is important
to point-out that, within this framework, energy and related
technological transitions have two ways to inﬂuence aggregate
transport demand (Fouquet, 2010): ﬁrst, by reducing the costs or
prices of transport services, which did occur in past transitions, and
can be represented by a slide down the demand curve; and, second,
through providing higher quality services, which also occurred, and
might be reﬂected by a shift in the demand curve.
Here, it is proposed that the dramatic increase in energy
consumption that followed the transition from stage coaches
(and horse power) to railways (and coal) was due to very high
income and price elasticities. Prices fell substantially generating
greater transport use. Even more important seems to have been
the dramatic shift in the demand curve (reﬂected in the high
income elasticity). The transition to the internal combustion
engine (and petroleum products), buses and cars, led to a large
increase in energy consumption. However, income elasticities
were around unity and price elasticities were less than one. In
other words, the energy transition from coal to petroleum
boosted energy consumption relatively less than the one from
horse power to coal. Given past trends, prior expectations and the
tentative forecast above, it is tempting to conclude that any future
energy transition in the land transport sector will increase energy
consumption relatively less than past transitions.
Having said this, the demand for transport-related energy is
driven ﬁrst by the demand for transport and then by the relationship
between the technology for providing transport and the
amount of energy required (Small and van Dender, 2007). The
extent of the increase in energy consumption will depend on
(i) the decline in transport prices, (ii) transport price elasticities,
(iii) the change in fuel efﬁciency, (iv) transport income elasticity
and (v) improvements in the quality of the transport service
associated with the transition, which will affect the income
elasticity. The latter highlights an important feature of energy
transitions: by raising the quality of the service, as cars and
arguably planes did, they delay the decline in elasticities. So,
when comparing two scenarios of the future (one with and one
without transition), unless the transition is towards a highly
energy efﬁcient technology (and this is possible), it is likely that
the scenario with an energy transition will generate higher energy
consumption. Whether this scenario produces more or less
carbon dioxide emissions will depend on the energy source used.
Up to this point, the discussion has been about behaviour in
the United Kingdom, and possibly by analogy in other developed
economies. Since this study looks at trends in elasticities at
different phases of economic development, it can provide some
clues to trends in developing economies. Their elasticities are
probably higher than in developed economies. So, as these economies
develop and incomes rise, transport use is likely to rise more
than proportionally, with major effects on the demand for current
energy sources and on the local and global environment.
It is possible that income and price elasticities are not as high
in developing economies today as in Victorian Britain – for a given
level of income, transport prices are substantially lower in
developing economies today than in nineteenth century England.
Also, as these economies develop, their elasticities will probably
also fall. Again, the introduction of new and superior quality
transport services associated with a transition will probably delay
or slow-down the declining trend.
This study focussed mostly on land transport. It appears that
elasticity estimates for air travel in developed economies are still
high. Thus, a technological and/or energy transition in air transport
services may well generate high increases in transport and
energy consumption in both developed and developing economies.
Again, the environmental impact will depend on the energy
source used after the transition.
This discussion implies that trends in energy consumption are
likely to continue to grow. There may be policies and strategies
that can reduce or stabilise transport demand, if that is desirable
(Bannister, 2011). However, the underlying argument of this
paper is that energy use is driven by the demand for mobility,
and energy and technological transitions will not stop this
demand or its trend. Nevertheless, this paper proposes that,
although future transitions will probably not generate the dramatic
increases in energy consumption experienced during transitions
in the nineteenth century, they may well alter (and shift
upwards) the trends in transport and related energy use, with
substantial implications for carbon dioxide emissions. Especially
given the relatively low price elasticity of demand for land
transport in industrialised countries, the role of policies, such as
carbon pricing or taxing, will be not so much to reduce demand
for travel but to encourage a shift towards low carbon energy
sources, technologies and possibly behaviour.
Acknowledgements
I would like to thank Bill Nordhaus, David Stern, Lester Hunt,
Lutz Killian, Peter Pearson, Paul Warde and Steve Sorrell, as well
as two anonymous referees, for comments related to this research
line and/or this paper. Of course, the author of this paper is solely
responsible for any errors.
References
Allen, R.C., 2007. Pessimism preserved: real wages in the British Industrial
Revolution. Economics Series Working Papers 314. University of Oxford.
Department of Economics. /http://www.nufﬁeld.ox.ac.uk/General/Members/
allen.aspxS.
Ayres, R.U., 2005. Resources, scarcity, technology and growth. In: Simpson, R.D.,
Toman, M.A., Ayres, R.U. (Eds.), Scarcity and Growth Revisited: Natural
Resources and the Environment in the New Millennium. Resources for the
Future, Washington D.C.
Bagwell, P.S., 1974. The Transport Revolution from 1770. B.T. Batsford, London.
Bannister, D., 2011. The trilogy of distance, speed and time. Journal of Transport
Geography 19 (4), 950–959.
Bentzen, 1994. J. Bentzen, An empirical analysis of gasoline demand in Denmark
using cointegration techniques. Energy Economics 16 (2), 139–143.
Bowden, S., Offer, A., 1994. Household appliances and the use of time: the United
States and Britain since the 1920s. Economic History Review 47 (4), 725–748.
BPP: British Parliamentary Papers, 1926. Coal Statistical Tables. Irish University
Press, Shannon, Ireland.
Broadberry, S. Campbell, B., Klein, A., Overton, M., van Leeuwen, B., 2009. British
Economic Growth, 1270-1870, Working Paper. /http://www2.lse.ac.uk/eco
nomicHistory/pdf/Broadberry/BritishGDPLongRun.pdfS.
Burnett, J., 1986. A Social History of Housing 1815–1985. Methuen, London.
Chartres, J., Turnbull, G., 1983. Road transport. In: Aldcroft, D., Freeman, M. (Eds.),
Transport in the Industrial Revolution. Manchester University Press,
Manchester.
Demographia, 2011. London Urbanized Area: Historical Estimated Population &
Density. /http://www.demographia.com/db-lonuza1680.htmS.
DoT, 2002. GB Transport Statistics (GBTS). HMSO, London.
DfT, 2011. GB Transport Statistics (GBTS). HMSO, London.
DTI, 1997. Digest of United Kingdom Energy Statistics (DUKES). HMSO. London.
DTI, 2001. Digest of United Kingdom Energy Statistics (DUKES). HMSO. London.
DTI, 2007. Digest of United Kingdom Energy Statistics (DUKES). HMSO. London.
DTI – Department of Trade and Industry, 2011 and backdates. Digest of United
Kingdom Energy Statistics. HMSO, London.
Elliott, G., Rothenberg, T.J., Stock, J.H., 1996. Efﬁcient tests for an autoregressive
unit root. Econometrica 64, 813–836.
R. Fouquet / Energy Policy 50 (2012) 62–7170
Foreman-Peck, J., 1981. The effect of market failure on the British motor industry
before 1939. Explorations in Economic History 18 (3), 257–289.
Fouquet, R., 2008. Heat Power and Light: Revolutions in Energy Services. Edward
Elgar Publications, Cheltenham and Northampton, MA, USA.
Fouquet, R., 2010. The slow search for solutions: lessons from historical energy
transitions by sector and service. Energy Policy 38 (10), 6586–6596.
Fouquet, R., 2011. Divergences in long run trends in the prices of energy and
energy services. Review of Environmental Economics and Policy 5 (2),
196–218.
Fouquet, R., Hawdon, D., Pearson, P.J.G., Robinson, C., Stevens., P.G., 1997. The
future of UK ﬁnal user energy demand. Energy Policy 25 (2), 231–240.
Fouquet, R., Pearson, P.J.G., 2012. The long run demand for lighting: elasticities and
rebound effects in different phases of economic development. Economics of
Energy and Environmental Policy 1 (1), 83–100.
Goodwin, P., Dargay, J., Hanly, M., 2004. Elasticities of road trafﬁc and fuel
consumption with respect to price and income: a review. Transport Reviews
24 (3), 275–292.
Greening, L.A., Greene, D.L., Diﬁglio, C., 2000. Energy efﬁciency and consumption –
the rebound effect – a survey. Energy Policy 28 (6–7), 389–401.
Gr¨ubler, A., 2004. Transitions in energy use. Encyclopedia of Energy 6, 163–177.
Gr¨ubler, A., Nakicenovic, N., Victor, D.G., 1999. Dynamics of energy technologies
and global change. Energy Policy 27, 247–280.
Hanly, M., Dargay, J., Goodwin, P., 2002. Review of income and price elasticities in
the demand for road trafﬁc, Report 2002/13. ESRC Transport Studies Unit,
University College London. London.
Hart, H.W., 1960. Some notes on coach travel. Journal of Transport History 4 (3),
146–160.
Hawke, G.R., 1970. Railways and Economic Growth in England and Wales (1840–
1870). Clarendon Press, Oxford.
Howarth, R.B., 1997. Energy efﬁciency and economic growth. Contemporary
Economic Policy 15 (4), 1–9.
Hughes, J.P., Knittel, C.R., Sperling, D., 2006. Evidence of a shift in the short-run
price elasticity of gasoline demand. NBER Working Paper No. 12530. National
Bureau of Economic Research.
Institute of Petroleum, 1994. UK Petrol Prices (1902–1993). Library & Information
Service. Institute of Petroleum, London.
Jackson, A.A., 2003. The London railway suburb, 1850–1914. In: Evans, A.K.B.,
Gough, J.V. (Eds.), The Impact of the Railway on Society in Britain: Essays in
Honour of Jack Simmons. Ashgate, Adershot.
Jackman, W.T, 1960. The Development of Transportation in Modern England.
Frank Cass. London.
Jevons, W.S., 1865. The Coal Question: An Inquiry Concerning the Progress of the
Nation, and the Probable Exhaustion of Our Coal-Mines. Macmillan, London.
Johansen, S., 1988. Statistical analysis of cointegration vectors. Journal of Economic
Dynamics and Control 12, 231–254.
Johansen, S., 1995. Likelihood-Based Inference in Cointegrated Vector Autoregressive
Models. Oxford University Press, Oxford.
Kilian, L., Murphy, D., 2010. The Role of Inventories and Speculative Trading in the
Global Market for Crude Oil, CEPR Discussion Papers 7753.
King, A.L., 1952. Statistics relating to the petroleum industry, with particular
reference to the United Kingdom. Journal of the Royal Statistical Society 115
(4), 534–565.
Leunig, T., 2006. Time is money: a re-assessment of the passenger social savings
from Victorian British railways. Journal of Economic History 66 (3), 635–673.
Luckin, B., 2000. The pollution in the city. In: Daunton, M.J. (Ed.), The Cambridge
Urban History of Britain: 1840–1950. Cambridge University Press, Cambridge.
Miller, K., 1993. Transport Sector Energy Model. Report for Economics and
Statistics Branch. Department of Trade and Industry, London.
Mitchell, B.R., 1988. British Historical Statistics. Cambridge University Press,
Cambridge.
Mok, D., Wellman, B., Carrasco, J.A., 2010. Does distance matter in the age of the
internet? Urban Studies 47 (13), 2747–2784.
MOP – Ministry of Power, 1961. Statistical Digest 1960. HMSO, London.
Moneta, A., Chai, A., 2010. The evolution of Engel curves and its implications for
structural change. Discussion Papers in Economics 201009. Grifﬁth University,
Department of Accounting, Finance and Economics.
Munby, D.L., 1978. In: Watson, A.H. (Ed.), Inland Transport Statistics: Great Britain,
1900–70, Vol. I. Clarendon Press, Oxford.
Nielsen, B., 2001. Order determination in general vector autoregressions. Working
Paper. Department of Economics. University of Oxford and Nufﬁeld College.
O’Connell, S., 1998. The Car in British Society: Class, Gender and Motoring.
Manchester University Press, Manchester.
ONS: Ofﬁce of National Statistics, 2002. Social Trends. HMSO, London.
Oum, T.H., Waters, II, W.G., Yong, J.S., 1990. A survey of recent estimates of price
elasticities of demand for transport. Working Paper. Policy, Planning and
Research. Infrastructure and Urban Development Department. The World
Bank.
Oum, T.H., Waters II, W.G., Yong, J.S., 1992. Concepts of price elasticities of
transport demand and recent empirical estimates. Journal of Transport
Economics and Policy 26 (2), 139–154.
Ramanathan, 1999. R. Ramanathan, Short- and long-run elasticities of gasoline
demand in India: an empirical analysis using cointegration techniques. Energy
Economics 21, 321–333.
Small, K.A., van Dender, K., 2007. Fuel efﬁciency and motor vehicle travel: the
declining rebound effect. The Energy Journal 28 (1), 25–52.
Salomon, I., 1985. Telecommunications and travel: substitutability or modiﬁed
mobility? Journal of Transport Economics and Policy 19 (3), 219–235.
Sch¨afer, A., 2000. Regularities in travel demand: an international perspective.
Journal of Transportation and Statistics 3 (3), 1–32.
Sch¨afer, A., Victor, D.G., 2000. Transportation Research Part A: Policy and Practice
34 (3), 171–205.
Selvanathan, E.A., Selvanathan, S., 1994. The demand for transport and communication
in the United Kingdom and Australia. Transportation Research (Part
B) 28 (1), 1–9.
Sorrell, S., 2007. The Rebound Effect: An Assessment of the Evidence for EconomyWide
Energy Savings from Improved Energy Efﬁciency. UK Energy Research
Centre, London.
Thompson, F.M.L., 1976. Nineteenth century horse-sense. Economic History
Review 39 (1), 60–81.
R. Fouquet / Energy Policy 50 (2012) 62–71 71