North versus South. Energy transition and intensity in Europe over ...
Transcript of North versus South. Energy transition and intensity in Europe over ...
Paper for session 49 Helsinki 2006
North versus South
Energy Transition and Energy Intensity in Europe
over 200 Years
Abstract
The paper examines energy consumption in Sweden, Holland, Italy
and Spain over 200 years, including both traditional and modern energy car-
riers. The article is based on totally new series of energy consumption in-
cluding traditional carriers along with modern sources. Our main purposes
are a closer examination of the process of the energy transition in Europe
and a revison of the prevailing idea of there being an inverted U-curve (or
Environmental Kuznets Curve –EKC-) in the very long run. Changes in en-
ergy consumption are decomposed into effects from population growth, eco-
nomic growth and energy intensity. The results on energy intensity challenge
previous suggestions by most scholars. An inverted U curve does not exist
whenever we include traditional sources of energy in our analysis.
Keywords: energy, traditional sources, EKC, service transition
Ben Gales, Astrid Kander, Paolo Malanima and Mar Rubio
Ben Gales Paolo Malanima
Faculty of Economics Institute of Studies
Economic History on Mediterranean Societies
University of Groningen (Italian National Council
The Netherlands of Research) Italy
[email protected] [email protected]
Astrid Kander* Mar Rubio
Department of Economic History Department of Economics
Box 7083 and Business
220 07 LUND University Pompeu Fabra
Sweden Barcelona Spain
[email protected] [email protected]
2
There is no consensus on precisely how important energy is for eco-
nomic growth and human welfare. If energy is a crucial resource for
the economy, so that growth can not take place without more or less
proportionate increases of energy, its availability could endanger
economic behaviour in a near future. If, on the other hand, energy
consumption shows high variability in relation to the economy, and
especially if it tends to decline in relative terms, the prospects are
more optimistic.
The proposition that energy consumption does not grow propor-
tionately to GDP, but that the ratio between energy and GDP (energy
intensity) shows a pattern that resembles an inverted U-curve, with a
relative increase in the early phases of industrialization and subse-
quently a decline in the post-industrial phase, is maintained by most
scholars.1 Nevertheless, this widely assumed proposition does not
take into account traditional forms of energy exploited prior to the in-
troduction of fossil fuels. It is our opinion that the long-term pattern
would look profoundly different from an inverted U-curve when tradi-
tional energy carriers, such as firewood and muscle energy of men
and working animals, were included in the calculation. In this case, in
fact, the starting level of energy use would not be so low: part of the
increase of energy consumption would appear, as a consequence,
only as a substitution of modern to traditional energy carriers.2 The
hypothesis was confirmed for Sweden, where there was a long-term
1 Goldenberg, Reddy (1990). 2 Jean-Marie Martin (1988) had suggested, in the wake of Schurr & Netschert (1960), that including firewood did change the energy/GDP relationship for the US, which was a wood-rich country. Grubler, A. (2004) reproduces these results for the US.
3
declining trend in energy intensity (energy measured by its heat con-
tent divided by GDP in constant prices), around which there were
some long swings.3 The result thus conveyed a fairly optimistic mes-
sage: economic growth has taken place with relatively less energy
requirements during the past 200 years; that is the economy man-
aged to grow faster than the consumption of energy. Furthermore, at
least in Sweden, a substantial relative decline took place from the en-
ergy crisis of the 1970s, after which even the absolute energy con-
sumption levelled off and came to a halt. Further reductions could be
expected in the future.
The Swedish long-term decline in energy intensity does not seem
apparently country-specific. A similar trend has been recently found
out in Italy, where energy intensity declined since the middle of the
19th century.4 So it seems reasonable to suppose that we could dis-
cover a similar decreasing trend of the energy intensity in other Euro-
pean countries.
The following research is based on new statistical material and
new series of energy consumption in four European countries.5
In section 1 we account for the different energy systems of our
four countries.6 We are dealing with two rather small, but early indus-
trialized and climatically cold countries in the North -The Netherlands
and Sweden-, versus two large but late industrialized Mediterranean 3 Kander (2002), Kander & Lindmark (2004). 4 Malanima (2006). 5 These data, together with the criteria used are available in www. (to be added). See also Kander (2002) and Kander & Lindmark (2004), Rubio (2005) and Malanima (2006). 6 Subsequently also Germany, England and Wales, Portugal and Norway will be in-cluded among the countries that we compare. Work on these countries is ongoing by researchers in our EGP (Energy-Growth-Pollution) network.
4
countries -Italy and Spain-. These four countries represent regions of
Europe with differences in climate, domestic sources of energy and
economic development paths.
In section 2 we compare the aggregate and per capita energy
consumption of the various countries. We find pair-wise similar results
in per capita consumption for the North and the South countries.
In section 3 we analyze the changes over time in aggregate en-
ergy consumption. For this purpose we use the Commoner-Ehrlich
equation and decompose the changes into population effects, con-
sumption effects and a residual.7 The residual is energy intensity (en-
ergy/GDP). We then will shortly address the discussion on the role of
the transition to the service economy, which we argue is largely an
illusion in terms of real production.
Section 4 contains a concluding discussion, which sums up the
main results of the study and makes some overall comments on what
has to be learnt from the previous analysis on the energy-economy
relationship.
1. The energy transition. Comparisons of the national energy
systems
The common feature of all these four countries is that traditional
energy carriers made up a large part of energy consumption until
7 Commoner (1971a), (1971b), Ehrlich et al. (1977), Ehrlich and Ehrlich (1990).
5
rather late (Figures 1-4). If we classify muscle power of men and
working animals, firewood, wind and water and peat as traditional en-
ergy carriers, we find that their contribution to the total energy input
became less than 50 percent only from 1864 on in The Netherlands,
from the late 1920s in Sweden, and immediately before World War II
in Italy and Spain.
Figure 1: The Swedish energy transition 1800-2000
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%non-thermalelectricityspent pulpingliquordistrict heating
gas
peat
o il
coal
firewood
muscle
Note: District heating is hot wastewater from industrial processes that are
used for heating houses. This is rather common in Sweden. Spent pulping
liquor is an energy-rich waste product from the pulp manufacturing that is
used as fuel in the pulp industry.
Figure 2: The Spanish energy transition 1850-2000
6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%non-thermalelectricityNatural Gas
Oil
Coal
Direct water
Fuelwood
M uscle
Figure 3: The Italian energy transition 1861-2000
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100% non-thermoelectricitygas
o il
coal
water
wind
firewood
muscle
Figure 4: The Dutch energy transition 1800-2000
7
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%non-thermalelectricitygas
o il
coal
peat
wind
firewood
muscle
The relative importance of specific traditional energy carriers dif-
fered among our countries.8 In Sweden the dominant traditional en-
ergy carrier was firewood, with roughly 75 percent of energy in 1850,
and food and fodder making up the rest. Direct working water was
less than 1 percent. In 1850 Spain the situation was pretty much the
reverse: food and fodder made up 50 percent, firewood 46, coal 1.7
and direct water and wind 2. This means that Spain had relatively
more motive power and less thermal energy at its disposal than Swe-
den. In 1861 Italy the situation reminded of the Spanish, but food for
men and fodder for draught animals played a minor role (about 20
percent each), firewood made up 50 percent and coal 7, while wind
8 The figures presented in these pages have been worked out using the same meth-odology. These methods are fully explained in Kander (2002) and Malanima (2006). While muscle energy corresponds to the input of food by men and working animals –only working animals -; firewood to the actual consumption in the four countries; wind and water have been estimated from the power of water and wind engines –sailships included- and the time, per year, these engines were in use. On modern carriers we exploited the statistical national accounts; or OECD and IEA series (in The Nether-lands from 1950 on).
8
and water only 1 percent.9 In Holland the situation was remarkable in
terms of the relative roles played by wind10 and peat. Around 1850
peat represented 25 percent and wind 10! The large pumping and
drainage projects needed in a country with half of its surface close to
or below the sea level were the main reason for the spectacular high
wind consumption. Firewood was, on the contrary, not very important
there: it only made up 10 percent, coal 15 and food and fodder 3511
(Table 1).
Table 1. Composition of energy consumption in 1850 Sweden, The Nether-
lands, Italy* and Spain (%).
Sweden Netherlands Italy Spain
Muscle 25 38 41 50
Firewood 73 11 51 46
Wind, Water <1 10 1 2
Fossil fuels 2 41 7 2
* 1861.
The coal-age arrived late in these countries, even though the age
of fossil fuels was already in progress in Holland, because of the wide
usage of peat (Table 2). Coal was very much a phenomenon of the
20th century.
9 Malanima (2006) and Bardini (1998). 10 Albers (2002), pp. 103-14 and 200-02. 11 Van Zanden (1997), p. 491. See also the remarks on firewood consumption in The Netherlands by Van der Woude (2003).
9
Table 2. Composition of energy consumption in 1900 Sweden, The Nether-
lands, Spain and Italy (%).
Sweden Netherlands Italy Spain
Muscle 17 20 39 31
Firewood 45 1 34 26
Wind, Water <1 3 1 5
Fossil fuels 38 76 26 38
Primary electricity 0.1 0 0.1 0
Note: primary electricity is only a commonly used abridged expression for
hydro- and nuclear electricity. Actually primary electricity does not exist,
electricity being in any case a secondary form of energy. Electricity is here
calculated by its heat content, and not by the energy content of the water or
uranium used for its production.
The dominance of the major energy carrier varied considerably.
The share of coal at its peak was 79 percent in The Netherlands
(1913), but only 45 percent in Sweden (1909). In Italy the maximum
was around 40 percent in 1935-40 and in Spain it peaked in the years
1927-30 with levels of 46-49 percent. Then the Civil War disrupted
the economy to such a degree that its effects could be noticed for the
fifteen years after its end. Similar levels of coal consumption were not
achieved again until the mid 1950s (Table 3).
Table 3. Composition of energy consumption in 1950 Sweden, The Nether-
lands, Spain and Italy (%).
10
Sweden Netherlands Italy Spain
Muscle 6 10 27 27
Firewood 21 0 17 12
Wind, Water <1 0 0 0
Fossil fuels 64 90 47 59
Primary electricity 9 0 10 2
The fact that The Netherlands became more of a coal econ-
omy than the others is not due to its own stocks.12 The take-off of in-
digenous mining was rather late, from 1900, and essentially from the
First World War onwards. In normal conditions this industry exported
most of the coal mined. Dutch demand, however, could be easily and
cheaply met by foreign trade.
The dependence upon oil was more similar across the coun-
tries. The break-through of oil came on grand scale after World War II
in Italy, Spain and Sweden, but less so in The Netherlands, which
remained coal based for a longer period (Table 4).
Table 4. Composition of energy consumption in 2000 Sweden, The Nether-
lands, Italy and Spain (%).
Sweden Netherlands Italy Spain
Muscle 2 2 4 4
Firewood 23 0 2 0
Fossil fuels 40 88 88 88
12 Gales (2000).
11
Primary electricity 33 10 6 7
Note: In 2000 Sweden had 2% of its energy supplied by district heating,
therefore the figures for Sweden do not sum up to 100%! Primary electricity
in The Netherlands is the sum of biomass electricity (7%), international trade
in electricity (more than 2%) and some nuclear electricity.
Primary electricity is of different importance in our four coun-
tries. It long was negligible in The Netherlands because of lack of wa-
ter falls, but nowadays make up 10 percent, where two thirds come
from bio-fuels and 1/3 from nuclear power. In Spain its importance in-
creased over time and is now 7 percent. In Italy the relative share de-
clined since its maximum in the 1950s (9-10 percent) and is nowa-
days 5 percent. Nearly all of this primary electricity is produced by
hydropower. Nuclear energy has been negligible in Italy and nowa-
days no nuclear power plant exists.13 Sweden is the country that has
put most effort into the primary electricity production. It presently con-
stitutes around 30 percent of the energy, half of which comes from
nuclear energy and half from hydropower. Sweden presently has 10
nuclear reactors running, out of originally 12. On the other hand, and
as a consequence of the large non-thermal electricity share, it has a
negligible share of thermo-electricity; only around 5 percent is pro-
duced by coal nowadays.
In 200 years, at the end of the 20th century, a long energy
transition had been accomplished with the almost total disappearance
13 Silvestri (1989), p. 175.
12
of traditional carriers such as wind, water and draft animals. Another
traditional source, human muscle energy, was relatively unimportant
from the start. Firewood, however, remained substantial in many
countries till deep in the 20th century. The more recent decades, after
the oil crises of the 1970s, show a proliferation of relevant energy car-
riers. The portfolio is less determined by one prime energy carrier
than in the past. There is, however, quite some variation between the
portfolios. The Netherlands became a gas-country. In Italy as well gas
consumption progressed fast since the oil crisis. Sweden opted for
nuclear electricity and, again, wood. Spain in turn, went back to coal
burning for electricity production. In the year 2000, about 90 per cent
of the coal consumed in Spain was for thermal electricity production.
2. Aggregate energy consumption
While structures of the energy systems are different, there are
clear similarities in the long-run patterns of total energy consumption
in the four countries (Figure 5). There were modest rates of increase
until the World War II, a period of faster growth rates in 1950-1970
and declining growth rates between 1970 and 2000. This fits well with
overall economic growth patterns and lend some credibility to the idea
of more growth requiring more energy and more energy making more
growth possible. Some differences are also obvious. First: the re-
sponse to crises such as the World Wars differed. The First World
13
War resulted in pair-wise similar effects: The Netherlands and Swe-
den decreased their energy consumption relatively much during that
War, while Spain and Italy show more modest declines. Spain had its
own civil war to cope with in the 1930s and for those years statistics
are lacking. Second: the interwar period was marked by strong
growth in energy consumption in The Netherlands and Sweden, while
energy requirements in Spain (who did not participate in the Secon
World War) and in Italy hardly grew at all. Third: Spain shows a
stronger growth in energy during the 1990s than the other countries,
which continue to have modest growth rates.
Figure 5: Primary energy consumption in Italy, the Netherlands,
Spain and Sweden, in petajoules (PJ).
Primary energy consumption, el (heat)
10
100
1000
10000
1800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
PJ
NetherlandsSwedenItalySpain
To some degree the different levels of energy consumption are
naturally connected to the size of the country, or basically to the
number of inhabitants of each country. The levels of energy con-
14
sumption, and the ordinal ranking of the countries in this respect,
show that Italy is the clear leader. Spain is second, but challenged in
that position by The Netherlands that catches up with Spain in the
1930s and then stays very close. Sweden starts at a level higher than
The Netherlands, but converges and reaches the same level in the
1880s, only to fall behind The Netherlands again from the 1920s on-
ward.
When examining the energy consumption per capita (Figure 6),
we see more distinct pair-wise developments in the North and South
European countries, together with a completely different hierarchy:
Sweden and Holland clearly overtake Italy and Spain. The Nether-
lands and Sweden also show remarkable resemblance in their long-
term development, apart from the period before 1870, where Sweden
actually declines its per capita energy consumption thanks to impres-
sive thermal efficiency developments both in household heating and
the iron industry, and in cultivation of new land without proportionate
increases in draught animals. Their development is thus one of con-
vergence. Spain and Italy share almost the same level of energy per
capita over the first 100 years (1860-1960), but then Italy takes off
and surpasses Spain. We see, however, that, while in the long term
climatic conditions, still important, count relatively less and the eco-
nomic performance is more meaningful, a relative convergence takes
place. The difference between Sweden-Holland on one hand and It-
aly-Spain on the other is much lower today than immediately after the
Second World War and during the previous century.
15
Figure 6: Energy consumption per capita
Energy/cap
10
100
1000
1800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
GJ/
cap
NetherlandsSwedenItalySpain
3. Decomposing energy
A frequently used formula for decomposing the environmental
impact into its main components is the Commoner-Ehrlich equation:14
I P C T= ⋅ ⋅
Where:
I environmental impact;
P population;
C per capita GDP;
T the environmental impact from producing one unit of GDP (the ratio
energy/GDP), which depends on the level of technology.
14For an overview, see Ekins (2000), pp. 154-81.
16
A growth in P and C needs to be outbalanced by declines in T
in order for I to stay constant. Technical change can imply shifts in
production possibilities so that the environmental stress per produced
unit is reduced. In other words: there is an ongoing race between
productivity increases and growth and the winner determines the rela-
tionship between economy and environment.15
Translated into an energy context, the Commoner-Ehrlich
equation gets the following expression, where E stands for total en-
ergy consumption, GDP for C, while the T-factor has been replaced
by energy intensity (E/GDP):
GDP EE PP GDP
= ⋅ ⋅
A look at their rate of growth during the half century 1950-
2000 and from the 1973 oil crisis until 2000 clearly reveals that the
forces of growth –population and GDP per capita- have been stronger
than the forces of the efficiency –energy intensity-. In the following ta-
ble e, p, y and ei are the annual compound rates of increase of total
energy consumption, population, per capita GDP and energy inten-
sity. Whenever p+y exceeds ei, the consequence is an increase of to-
tal energy consumption (Table 5).
Table 5. Compound yearly growth rates in Energy (total), Population, per
capita GDP, Energy intensity (1950-2000 and 1973-2000)(%).
15 Grubler (1998).
17
Sweden Netherlands Italy Spain
1950-
2000
1973-
2000
1950-
2000
1973-
2000
1950-
2000
1973-
2000
1950-
2000
1973-
2000
e 1.92 0.12 3.08 1.07 3.80 0.85 3.80 2.65
p 0.46 0.32 0.91 0.63 0.43 0.27 0.76 0.60
y 2.20 1,45 2.59 2,50 3.62 2.02 3.91 2.50
ei -0.73 -1.65 -0.42 -1.40 -0.27 -1.43 -0.87 -0.45
We see, however, that when we distinguish the period 1973-
2000 within this half century, the forces of productivity and efficiency
reveal their stronger impact on the overall trend of energy consump-
tion. They have been able to deeply reduce the growth of energy
consumption and finally, at the very end of the century, to neutralize
it. In 1973-2000, energy intensity decline was faster than per capita
GDP and population growth, taken separately. In any case, taking the
27 years between 1973 and 2000 into account, the combined effect of
p+y, has been stronger than the decline in ei .The whole result – the
slower increase of e than before- has been more easily attained
thanks to the slower growth rate of GDP and population.
A disaggregate analysis per country and per factor provides a
clearer picture of the overall trend.
3.1 The P-factor
The P-factor, or the population factor (Table 6), explains why
The Netherlands overtook Sweden in energy consumption in the
18
1920s: after this point in time the growth rate of the Dutch population
was much higher than that of the Swedish. The Netherlands were a
demographic outlier for a considerable time. The number of births
remained high after the demographic transition in combination with a
very low rate of female participation in the labour force. The latter only
converged to the West-European pattern in the 1970’s; the former
had come down somewhat earlier. The factors behind this were and
still are a matter of debate, but cultural and religious forces are hold-
ing up as important.
Table 6. Population in Sweden, The Netherlands, Italy and Spain 1800-
2000 (thousands).
Sweden Netherlands Italy Spain
1800 2,336 2,112 18,260 10,392*
1825 2,771 2,514 20,134 12,615
1850 3,482 3,098 24,603 14,894
1875 4,383 3,788 28,258 16,267
1900 5,136 5,133 33,343 18,594
1925 6,053 7,362 38,715 22,433
1950 7,041 10,113 46,768 27,976
1975 8,208 13,666 54,764 35,548
2000 8,882 15,925 57,844 40,933
* 1797
This P-factor shows small variations in growth rates over time
(between 0.40 and 0.80 percent per year in 1950-2000), and is thus
19
incapable of explaining a factor that shows large fluctuations, such as
energy. For this we need to look at the income per capita variable, the
C-factor.
3.2 The C-factor
The C-factor or the income per capita (Figure 7) is a more
volatile variable than the P-factor and as such is more able to “ex-
plain” development of the E-factor over time. Energy consumption,
however, is not merely a function of income, as the differences be-
tween countries show.
Income per capita
100
1000
10000
100000
1800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
inte
rnat
iona
l dol
lars
NetherlandsSwedenItalySpain
Figure 7: Income per capita, constant international 1990 PPP dollars
Naturally this is not the place to discuss the long-term economic
development of these four countries, but the differences across them
appear as much in levels as in trends. The start of Modern Growth
coincided in these four countries with the passage to the exploitation
20
of modern energy carriers and with the transition from an energy sys-
tem based on vegetable carriers to a mineral one. For a long time
The Netherlands enjoyed a leading position. When, after the World
War II, GDP growth accelerated, a trend toward convergence charac-
terized the four economies. At the end of the century, the levels of per
capita GDP are the same in Sweden, Holland and Italy, and Spain is
rapidly converging.
The rise in per capita GDP was the strongest variable in de-
igure 8: Energy per capita versus income per capita
3.3. The T-factor: Energy intensity
termining a unprecedented growth in energy consumption (Figure 8),
even though energy intensity decline contributed to partially neutralize
this upward trend. In log figures a nearly linear relationship exists be-
tween GDP and energy consumption
F
Energy per capita versus income per capita in several European countries (Netherlands, Sweden, Italy and, Spain 1850-2000)
10
100
1000
1.000 10.000 100.000GDP per habitant (log)
international dollars (1990$) per habitant
ener
gy p
er c
apita
(log
) G
J/ha
bita
nt y
ear
Sweden Netherlands Italy Spain
20.0005.000
Netherlands 1995Eng: 210 GJ/habGDPpc: 18527 $/hab Sweden 2000
Eng: 168 GJ/habGDPpc:19898 $/hab
Italy 2000Eng: 134 GJ/habGDPpc: 20574
Spain 2000Eng: 119 GJ/habGDPpc: 14927
21
T is a factor that by construction catches up everything that
has not been caught by the P- factor and C-factor. It stands here for
energy intensity, E/GDP, and is portrayed in figure 9.
One main result is that Sweden is an outlier in its energy in-
tensity. The long-term decline is impressive in Sweden. Some anal-
ogy exists with Italy, but in none of the other countries it was the
same as in Sweden. A weak linear decline can be discerned for
Spain, but The Netherlands show no time trend at all. The reason is
that, as an early comer, its energy system was already different and,
he three late comers. Still,
ith the exception of The Netherlands, the levels of energy intensity
s the end of the
0th ce
in a sense, more modern than those of t
w
in the 19th century are larger than in the 20th century. Despite the wide
increases in energy consumption over the last 100 years, producing
has slowly become cheaper in energy terms. Toward
2 ntury, three out of the four countries have come down to less
than 10MJ/$, a level which is quite below the levels during the previ-
ous 100 years.
Figure 9: Energy intensity
22
Energy intensities, MJ/dollar
100,00
NetherlandsSweden10,00ItalySpain
1,001800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Efficiency in energy consumption, the ratio, that is, of the out-
put of useful energy to the total input, certainly improved during the
transition. It was estimated that, while a subsistence agriculture ex-
ploits energy with an efficiency of about 10 percent, a more advanced
agrarian economy economy (an economy, that is, based on food, fod-
der and
plants as the main carriers) before industrialization can reach
even 25 percent.16 This is probably an overestimation. We know that
the efficiency of a working animal hardly reaches 10 percent. Most
fodder is used for the metabolism and does not produce mechanical
energy. In the case of a human being, efficiency is higher: about 20
percent. Traditional fireplaces and stoves usually had a very low effi-
ciency, hardly reaching 20 percent and in most cases even 10. A
weighed average provides an estimate of overall efficiency at around
20 percent and perhaps less. Today modern energy systems are
credited with an efficiency of about 35 percent. In the end, the pas-
sage to the use of modern carriers and the exploitation of the ma-
chines as converters resulted in a higher efficiency. Biological, animal
16 Cook (1976), p. 135. For a more in-depth discussion of this issue, see Malanima (1996), pp. 119 ff.
23
converters of traditional, vegetable sources are, on the whole, less
efficient than the inanimate machines.
The transition from traditional energy carriers to modern ones
therefore implied a decrease of energy intensity in any country. At the
same time technological improvements in mechanical converters con-
tributed in that the declining trend continued and even intensified.
How does our approach change the perception of the relation-
gy
ship between energy intensity and economic growth? The most
widely accepted perspective on this relationship is probably the one
by Goldenberg and Reddy (1990), later identified as the so called En-
vironmental Kuznets Curve (Figure 10). According to this idea, ener
intensity will increase at low levels of income per capita, as the coun-
tries industrialize, and then, after attaining a certain level of per capita
income, energy intensity will start to decrease. A straight implication
of the Environmental Kuznets Curve is that developing nations could
achieve comparable levels of economic growth with a lower ratio of
consumed energy to GDP after a transition across a path of increas-
ing energy intensity.
Figure 10: Traditional portrait of the long-term evolution of energy in-
tensities
24
Source: adapted from Reddy, Goldenberg (1990).
rriers changes the im-
lied relation. The immediate consequence is to elevate the levels of
ensity and economic output per capita has not the expected
The inclusion of traditional energy ca
p
energy consumption, and thus energy intensities, of the earlier peri-
ods. In addition, the inclusion changes the shape. The Environmental
Kuznets Curve normally has the levels of income per capita rather
than time on the x-axis in contrast to the Reddy, Goldenberg presen-
tation and we follow that custom. In Figure 11 energy intensity in-
cludes all types of energies put into use for the economic system and
it is represented in the y-axis. Income per capita, measured by the
GDP per inhabitant of each of our four countries is represented in the
x-axis.
Our representation of the very long-run relationship between en-
ergy int
inverted-U shape implied by the Environmental Kuznets Curve. For
the most part, the figure looks more like a sea-wave moving towards
25
higher levels of income (to the right) with a long-term trend of lower
levels of energy intensity. In fact, the only inverted-U shape that can
be identified is the one peaking around the first oil crisis for three of
the four countries (Spain being the exception). From the level $8,000
per capita (obtained around 1960 in Italy, Sweden and The Nether-
lands) the levels of energy intensity increased until the level of
$13,000 per capita (which were achieved in 1973 by all three). From
then on all three countries achieved higher levels of income per cap-
ita with lower levels of energy intensity. The decline in energy inten-
sity caused by the oil crisis is also observable in the Spanish case,
but at a much lower level of income per capita: $7,700.
Figure 11: Energy intensity versus income per capita
Total energy intensity versus income per capita in several European countries (Netherlands, Sweden, Italy and, Spain 1850-2000)
1,0
10,0
1.000 10.000 100.000GDP per habitant (log)
international dollars (1990$) per habitant
ener
gy in
tens
ity(lo
g)
MJ
of e
nerg
y pe
r $ p
rodu
ced
(199
0$)
100,0Sweden Netherlands Italy Spain
Sweden 1860GDPpc:$1,076Eng: 26.3 MJ/$
In the longer run, the trend exhibits a steady decline of the en-
ergy intensity at higher levels of income per inhabitant. Therefore,
what remains untouched of the original representation of the relation-
ship between energy intensity and income per capita proposed by
Goldenberg and Reddy is the fact that latecomers may achieve com-
Spain 1850GDPpc: $1,079Eng: 13.7MJ/$ Spain 2000
GDPpc: $14,927Eng: 8 MJ/$
Netherlands 1973
20.0005.000
Sweden 2000GDPpc: $19,898Eng: 8.2 MJ/$
Italy 2000GDPpc: $20,574Eng: 6.5 MJ/$
26
parable levels of economic output per capita with a lower ratio of con-
sumed energy to GDP.
3.4 Unpacking the T-factor
ensity, is a residual and may be
erceived as technical change in a broad sense. This technical
y services increases as they get relatively cheaper,
The T-factor, or energy int
p
change can both stimulate efficiency increases and structural change.
The efficiency can arise both from process innovations that increase
efficiency within a specific kind of production, from the introduction of
superior energy carriers, or from product innovations that give rise to
new ventures and branches, thus stimulating structural change of the
economy. Within one existing branch or sector the innovations can be
explicitly directed at energy savings (such as improving the thermal
efficiency of a machine) or at innovations that save on the production
factors that need energy to run (the total factor productivity). If we
consider the large improvements in thermal efficiency of machines
that have taken place over time, together with the substantial im-
provements in productivity, we may erroneously be led to expect a
very radical decline in energy intensity in the long run. However two
things complicate the picture and make the gains in energy intensity
less predictable:
1. “rebound” or “take-back” effects, which means that the con-
sumption of energ
27
which is the case when technical energy efficiency improves.17 This
stimulates structural change in a more energy intensive direction;
2. biases in technical change, which may lead to less gains in
energy savings than in total factor productivity. A normal feature of
y improvements of various kinds in the
tion and decline during the transition to the service econ-
omy.19
the economic development is that capital per worker (K/L) increases.
The capital in relation to GDP (the K/Y ratio) varies among countries..
In cases where capital is not saved upon in relation to GDP it may be
that energy is not saved either, although in general energy is saved
upon in relation to capital.18
Thus energy intensity may not decrease at all, even though
we have substantial efficienc
economy.
A widely held belief is that energy intensity will increase during
industrializa
We think there is reason to be sceptical about the idea that
the transition to the service economy will bring about any relative de-
cline in energy intensity.20 This is mainly because there has not been
a uniform pattern of substantial growth of the share for the service
sector in real terms (constant prices). Between 1970 and 2000 the
service sector only increased by 2 percentage units in Sweden, while
transports grew by 4 percentage units. In Spain the service sector ac-
tually declined by one percentage unit (transports included) and in
Holland services grew by 5 percentage units, while transports grew
17 Howarth (1997). 18 Kander and Schön (forthcoming); Ayers and Warr (2003); Ayres and Warr (2005). 19 Panayotou, Peterson and Sachs (2000). 20 Kander (2005).
28
by another 2 percentage units. Only Italy saw a relatively large in-
crease of the service sector of 14 percent units (transports included).
As a consequence the substantial decline in energy intensity after
1970 in these countries cannot be explained by a relative growth of
services. The shift to a service economy is largely an illusion in
terms of real production, since it is generated by the fall in the price of
manufacturing goods relative to services, which is in turn caused by
more rapid productivity growth in manufacturing than in services.21
What we can consume a lot of in the service economy is what we are
efficient in producing, which is still mainly industrial goods. It is of
course true that parts of the service sector have high productivity and
contain inputs from the industrial sector, like computers, which com-
plicates the picture.22 Baumol’s main argument is still relevant: that
for activities where human time makes up part of the product (the ser-
vice) it is not possible to rationalize production by substituting ma-
chines for labour, and such labour intensive activities make up a large
share of the service sector.
Still the more rapid decline in energy intensity after 1970 com-
pared to earlier periods is a fact and we believe this has much to do
with th
e structural changes within the industrial sector, with a relative
growth of energy-light branches like information and communication
technology and bio-technology. The grand transformation of societies
related to the micro-electronic revolution, sometimes called the third
21 Baumol (1967), Kander (2005). 22 Baumol (1987).
29
industrial revolution, may be largely responsible for the relative sav-
ings of energy.23
4. Concluding remarks
aims in the present research have been:
y transition
2. opinion that energy
We have
Swede
Our two
1. the analysis of the process of the energ
in four European countries, representative, so to
speak, of the North and South;
the revision of the widely held
intensity shows the pattern of an inverted U-curve.
investigated energy consumption in the long run in
n, Holland, Spain and Italy and the energy transition taking
place in our four countries. The relative importance of the traditional
energy carriers differed among the countries, with Sweden relying
upon firewood, Spain and Italy upon fodder and food, and wind and
peat being substantial in The Netherlands. The coal age was a 20th
century phenomenon, but also here its maximum share in energy
consumption varied between 42 and 79 percent. In order to analyse
energy consumption we used a Commoner-Ehrlich type of decompo-
sition of the aggregate into major components, population, income
per capita and energy intensity as a proxy for technology.
23 Schön, L. (1990), Kander (2005).
30
Our main result has been the reshaping of the trend of the en-
ergy intensity on the basis of our new data on traditional energy carri-
ers in a very long perspective. The dominating view of a turned U
movement of the energy intensity is not confirmed by our research
and depended on concentration, by most scholars, only upon ‘mod-
ern’ energy carriers. Our figures show, on the contrary, a long-term
decline in energy intensity. As to the causes of this declining trend,
there is reason to be sceptical about the commonly held belief of a
correlation between the energy intensity decline and the rising impor-
tance of services in the economic structure. Even though a closer in-
spection of this topic would require a reexamination of the national
accounts and the way services are computed, our opinion is that
technological changes accounted much more for this decline than the
rise of services. In any case, looking at the movement of the energy
consumption and energy-GDP in a very long perspective, there is
some ground to be less pessimistic than the majority of the observ-
ers. Technical advance in energy use has been until now a limiting
force on the rising path of energy consumption. Its importance has
grown in the last few decades. The increase in the productivity of en-
ergy could realistically be a strong support of the economic activity in
a near future as well.
31
REFERENCES
Albers R.M. (2002). Machinery investment and economic growth. The dy-
namics of Dutch development 1800-1913. Amsterdam: Aksant, 2002.
Arpi, G. (1959). Sveriges skogar under 100 år" del II: Stockholm.
Ayers, R. and Warr, B. (2003). Exergy, power and work in the US economy,
1900-1998, Energy 28: 219-73.
Ayres, R., Warr, B. (2005). Accounting for growth: the role of physical work
Structural Change and Economic Dynamics, vol 16, issue: 2, pp 181-209.
Bardini C. (1998). Senza carbone nell’età del vapore. Gli inizi dell’industria-
lizzazione italiana. Milano: B. Mondadori.
Bartoletto, S. (2004). “Dalla legna al carbon fossile. I consumi di
combustibile a Napoli nel corso dell’Ottocento”. Melanges de l’Ecole Fran-
çaise de Rome, 116: 705-21.
Collantes Gutiérrez, F., Vicente Pinilla Navarro. DT 2002-3, Centro de Estu-
dios sobre Despoblación y Desarrollo de Áreas Rurales
Commoner, B. (1971a). The Closing Circle: confronting the Environmental
Crisis, Jonathan Cape, London.
32
Commoner, B. (1971b). The Environmental Cost of Economic Growth.
Schurr, S. (ed.) Energy, Economic Growth and the Environment, Baltimo-
re/London: John Hopkins University Press: 30-65.
Cramér, M. (1991). Den verkliga kakelugnen. Stockholm.
Ehrlich, P. and Ehrlich, A. (1990). The Population Explosion. London: Hut-
chinson.
Ehrlich, P., Ehrlich, A. and Holdren, J. (1977). Ecoscience: Population, Re-
sources, Environment. W.H. Freeeman, San Francisco, CA.
Ekins, P. (2000). Economic Growth and Environmental Sustainability: The
Prospects for Green Growth. London: Routledge.
Erixon, S. ( ). ”Spjället, en exponent för svensk bostadsteknik”. Svenska
kulturbilder, ny följd, 5.
Federico, G. (2003). L'agricoltura italiana: successo o fallimento? Ciocca P.,
Toniolo G. (eds). Storia economica d'Italia, 3, Industrie, mercati istituzioni. 1
Le strutture dell'economia. Bari-Roma: Laterza: 71-98.
Fenoaltea, S. (2003). Lo sviluppo dell’industria dall’Unità alla Grande Gue-
rra: una sintesi provvisoria. Toniolo G., Ciocca P. (eds.), Storia economica
d’Italia, 3: Industrie, mercati, istituzioni, t.1, Le strutture dell’economia. Ro-
ma- Bari: Laterza: 137-93.
Gales, B.P.A. (2000). Delfstoffen. Schot, J.W. e.a (eds.), Techniek in Neder-
land in de twintigste eeuw, Deel II. Delfstoffen, Energie en chemie. Zutphen:
Walburg Pers: 17-111.
33
Goldenberg, J., Reddy, A.K.N. (1990). “Energy for the Developing World”.
Scientific American 263 (3): 111-18.
Grubler, A. (1998) Technology and Global Change, Cambridge University
Press, IIASA.
Grubler, A. (2004) Transitions in Energy Use, Encyclopedia of Energy, vol 6.
Howarth, R. (1997). “Energy Efficiency and Economic Growth”. Contempora-
ry Economic Policy (15): 1-9.
Kander, A. (2002). Economic Growth, Energy Consumption and CO2 Emis-
sions in Sweden 1800-2000. Lund: Lund Studies in Economic History (19).
Kander, A. (2005). “Baumol’s Disease and Dematerialization of the Econo-
my”. Ecological Economy 55: 119-30.
Kander, A., Lindmark M. (2004). “Energy consumption, pollutant emissions
and growth in the long run -Sweden through 200 years”. European Review
of Economic History (3): 297-335.
Kander, A., Schön, L. The energy-capital relation, Sweden 1870-2000.
Structural Change and Economic Dynamics (forthcoming).
Krantz, O. (1987a). Husligt arbete 1800-1980. Lund:
Krantz, O. (1987b). Offentlig verksamhet 1800-1980. Lund:
Krantz, O. (1991). Privata tjänster 1800-1980 . Lund:
Krantz, O. (1991). Transporter och kommunikationer 1800-1980. Lund:
1986.
34
Larsson, H. (1979). Vedeldning genom tiderna-från Cronstedts kakelugn till
Hugos kamin. Tekniska högskolornas energiarbetsgrupp. Rapport nr 5.
Lindahl, E., Dahlgren, E., Koch, K. (1937). Wages, Cost of Living and Na-
tional Income in Sweden 1860-1930. Stockholm: Institute for Social Sci-
ences, University of Stockholm.
Lindqvist, S. (1984). ”Naturresurser och teknik-energiteknisk debatt i Sverige
under 1700-talet”. Frängsmyr, T. (ed.): Paradiset och vildmarken. Stock-
holm:
Ljungberg, J. (1988). Deflatorer för industriproduktionen 1888-1955. Lund:
Maddison, A. (1995). Monitoring the World Economy, 1820-1992. Paris:
OECD.
Malanima, P. (1996). Energia e crescita nell’Europa preindustriale. Roma:
La Nuova Italia Scientifica.
Malanima, P. (2006). Energy Consumption in Italy in the 19th and 20th Centu-
ries. A Statistical outline. Roma: CNR.
Martin, J.-M. (1988). ‘L’intensité énergétique de l’activité économique dans
les pays industrialisés: les évolutions de trés longue période livrent-elles des
enseignemets utiles? Economies et Sociétés, Series Economie de L’Energie
(4): 9-27.
35
Nadal, J. (1975). El fracaso de la revolucion industrial en Espana, 1814-
1913. Barcelona: Ariel.
Nadal, J. (1978). La industrialización y el desarrollo- económico de España.
Barcelona: Universidad de Barcelona.
Panayotou, T., Peterson, A., Sachs, J. (2000). Is the environmental Kuznets
curve driven by structural change? What extended time series may imply for
developing countries. CAER II, Discusión Paper n.° 80.
Prados de la Escosura L. (2003). El progreso económico de España 1850-
2000. Madrid: Fundación BBVA.
Rubio, M.d.M., 'Economía, Energía y CO2: España 1850-2000', Cuadernos
economicos de ICE, No.70, (2005)
Schön, L. (1990). Elektricitetens betydelse för svensk industriell utveckling.
Stockholm: Vattenfall.
Schurr, S.H., Netschert B.C. (1960). Energy in the American Economy. Bal-
timore, Md.: Johns Hopkins University Press.
Silvestri, M. (1989). Il futuro dell’energia. Torino: Bollati Boringhieri.
Smits, J.-P., Horlings, E., Van Zanden, J.L. (2000). Dutch GNP and its Com-
ponents, 1800-1913. Groningen: Groningen Growth and Development Cen-
tre.
36
Sudrià, C. (1987). Un factor determinante: la energía. Nadal J., Carreras, A.,
Sudrià, C. (eds.) La economía española en el siglo xx. Una perspectiva his-
tórica. Barcelona: Ariel.
Sudrià, C. (1995). Energy as a Limiting Factor to Growth. Martín P.S, Aceña,
J. (eds.) The Economic Development of Spain since 1870. Aldershot: Ed-
ward Elgar.
Van der Woude, A. (2003). Sources of Energy in the Dutch Golden Age.
Cavaciocchi, S. (ed.). Economia e energia secc. XIII-XVIII. Firenze: Le
Monnier: 445-68.
Van Zanden, J.L. (1997). “Werd de gouden eeuw uit turf geboren? Over het
energieverbruik in de Republiek in de zeventiende en achttiende eeuw”.
Tijdschrift voor Geschiedenis (110): 484-99.
37