Energy and Civilization: A History by Vaclav Smil
My rating: 5 of 5 stars

Great Book, Smil manages to get the science across in an accessible style without utilising the memes of the Climate Change Cult. The Book is all the more serious and important because of that.
here is a Writer that should be taken seriously.
http://theconquestofdough.weebly.com/

View all my reviews

Box 1.10
Calculating the net energy cost of human labor
There is no universally accepted way to express the energy cost of human
labor, and calculating the net energy cost is perhaps the best choice: it is a
person’s energy consumption above the existential need that would have to be
satisfied even if no work were done. This approach debits human labor with
its actual incremental energy cost. Total energy expenditure is a product of
basal (or resting) metabolic rate and physical activity level (TEE = BMR × PAL),
and the incremental energy cost will obviously be the difference between TEE
and BMR. The BMR of an adult man weighing 70 kg would be about 7.5 MJ/d,
and for a 60 kg woman it would be about 5.5 MJ/day. If we assume that hard
work will raise the daily energy requirement by about 30%, then the net
energy cost would be about 2.2 MJ/day for men and 1.7 MJ/day for women,
and hence I will use 2 MJ/day in all approximate calculations of net daily
energy expenditures in foraging, traditional farming, and industrial work.
Daily food intake should not be counted as an energy input of labor: basal
metabolism (to support vital organs, circulate the blood, and maintain a
steady body temperature) operates regardless of whether we rest or work. Stud-
ies of muscle physiology, especially the work of Archibald V. Hill (1886–1977,
recipient of the Nobel Prize in Physiology in 1922), made it possible to quan-
tify the efficiency of muscular work (Hill 1922; Whipp and Wasserman 1969).
The net efficiency of steady aerobic performances is about 20%, and this
means that 2 MJ/day of metabolic energy attributable to a physical task would
produce useful work equal to about 400 kJ/day. I will use this approximation
in all relevant calculations. In contrast, Kander, Malanima, and Warde (2013)
used total food intake rather than actual useful energy expenditure in their
historical comparison of energy sources. They assumed an average annual
food intake of 3.9 GJ/capita, unchanged between 1800 and 2008.

Box 1.7

Comparison of energy returns in food production

Since the early 1970s, energy ratios have been used to illustrate the superiority

of traditional farming and the low energy returns of modern agriculture.

Such comparisons are misleading owing to a fundamental difference between

the two ratios. Those for traditional farming are simply quotients of the food

energy harvested in crops and the food and feed energy needed to produce

those harvests by deploying human and animal labor. In contrast, in modern

farming the denominator is composed overwhelmingly of nonrenewable fos-

sil fuel inputs needed to power field machinery and to make machines and

farm chemicals; labor inputs are negligible.

If the ratios were calculated merely as quotients of edible energy output to

labor input, then modern systems, with their miniscule amount of human

effort and with no draft animals, would look superior to any traditional prac-

tice. If the cost of producing a modern crop included all converted fossil fuels

and electricity converted to a common denominator, then the energy returns

in modern agriculture would be substantially below traditional returns. Such

a calculation is possible because of the physical equivalence of energies. Both

food and fuels can be expressed in identical units, but an obvious “apples and

oranges” problem remains: there is no satisfactory way to compare, simply

and directly, the energy returns of the two farming systems that depend on

two fundamentally different kinds of energy inputs

p.19-20
Chapter 1

A circle is closed. I have noted the necessity of quantitative evaluations,
but the real understanding of energy in history requires much more than
reducing everything to numerical accounts in joules and watts and treating
them as all-encompassing explanations. I will approach the challenge in
both ways: I will note energy and power requirements and densities and
point out improving efficiencies, but I will not ignore the many qualitative
attributes that constrain or promote specific energy uses. And while the
imperatives of energy needs and uses have left a powerful imprint on
history, many details, sequences, and consequences of these fundamental
evolutionary determinants can be explained only by referring to human
motivations and preferences, and by acknowledging those surprising, and
often seemingly inexplicable, choices that have shaped our civilization’s
history.

Energy in Prehistory
41
Before leaving the forager energetics I should note that foraging retained
an important role in all early agricultural societies. In Çatalhöyük, a large
Neolithic agricultural settlement on the Konya Plain dated to about 7200
BCE, early farmers had diets dominated by grains and wild plants, but
excavations also show the bones of hunted animals, ranging from large
aurochs to foxes, badgers, and hares (Atalay and Hastorf 2006). And at
Tell Abu Hureyra in northern Syria, hunting remained a critical source of
food for 1,000 years after the beginning of plant domestication (Legge and
Rowley-Conwy 1987). In predynastic Egypt (earlier than 3100 BCE), the
cultivation of emmer wheat and barley was complemented by the hunting
of waterfowl, antelopes, wild pigs, crocodiles, and elephants (Hartmann
1923; Janick 2002).
Origins of Agriculture
Why did some foragers start to farm? Why did these new practices diffuse
so widely, and why did their adoption proceed at what, in evolutionary
terms, is a fairly rapid rate? These challenging questions may be sidestepped
by agreeing with Rindos (1984) that agriculture has no single cause but
arose from a multitude of interdependent interactions. Or, as Bronson
(1977, 44) put it, “What we are dealing with is a complex, multifaceted
adaptive system, and in human adaptive systems … single all-efficient
‘causes’ cannot exist.” But many anthropologists, ecologists, and historians
have been trying to find precisely such principal causes, and there are many
publications surveying diverse explanatory theories about the origin of
agriculture (Cohen 1977; Pryor 1983; Rindos 1984; White and Denham
2006; Gehlsen 2009; Price and Bar-Yosef 2011).

There was no single center of domestication from which cultivated
plants and milk- and meat-producing animals spread, but in the Old World
the most important region of agricultural origin was not, as previously
thought, the southern Levant but rather the upper reaches of the Tigris and
Euphrates rivers (Zeder 2011). This means that food production started
along the margins, rather than in the core areas, of optimal zones. The
botanical record from Chogha Golan in the foothills of the Iranian Zagros
Mountains provides the most recent confirmation of this reality (Riehl,
Zeidi, and Conard 2013): cultivation of wild barley (Hordeum spontaneum)
began there about 11,500 years ago, later augmented by the cultivation of
wild wheat and wild lentils.
In process terms, it is essential to stress that there are no clear thresh-
olds or sharp divides between foraging and agriculture, as extended
periods of managing wild plants and animals precede their true domestica-
tion, which is characterized by clearly identifiable morphological changes.
And, contrary to earlier understandings, the domestication of plants
and animals proceeded almost concurrently and yielded an effective
This content downloaded from 139.184.14.159 on Fri, 23 Jun 2017 08:00:31 UTC
All use subject to http://about.jstor.org/termsEnergy in Prehistory
43
arrangement fairly rapidly (Zeder 2011). The oldest approximate dates for
the first appearance are about 11,500–10,000 years before the present for
the plant species emmer (Triticum dicoccum) and einkorn wheat (Triticum
monococcum) and barley (Hordeum vulgare) in the Middle East (fig. 2.4),
10,000 years for China’s millets (Setaria italica), 7,000 years for rice (Oryza
sativa), 10,000 years for Mexican squash (Cucurbita species), and 9,000
years for corn (Zea mays) and 7,000 years for Andean potatoes, Solanum
tuberosum (Price and Bar-Yosef 2011). The earliest animal domestications
go back to 10,500–9,000 years ago, starting with goats and sheep, followed
by cattle and pigs.
The two main explanations of Europe’s Neolithic transition to farming
have been through indigenous action animated by imitation (cultural dif-
fusion) or driven by dispersing populations (demic diffusion). Radiocarbon
dating of material from early Neolithic sites by Pinhasi, Fort, and Ammer-
man (2005) yielded results consistent with the prediction of demic diffu-
sion, radiating most likely from the northern Levant and the Mesopotamian
area and proceeding northwestward at an average pace of 0.6–1.1 km/year.
This conclusion is supported by comparisons of mitochondrial DNA
sequences from late European hunter-gatherer skeletons with those from

(a)

(b)

(c)

Figure 2.4
The earliest domesticated cereals. a–c. Emmer wheat (Triticum dicoccum), einkorn
wheat (Triticum monococcum), and barley (Hordeum vulgare) were the foundation of
the origins of agriculture in the Middle East (Corbis).

Two proteins in wheat are unique, not nutritionally but because of
their physical (viscoelastic) properties. Monomeric gluten proteins (gliadin)
are viscous; polymeric gluten proteins (glutenin) are elastic. When com-
bined with water they form a gluten complex that is sufficiently elastic to
allow a leavened dough to rise, yet strong enough to retain carbon dioxide
bubbles formed during the yeast fermentation (Veraverbeke and Delcour
2002).
Without these wheat proteins there would be no leavened bread, the
basic food of Western civilization. Yeast was never a problem: wild (natu-
rally occurring) Saccharomyces cerevisiae is present on the skins of many
fruits and berries, and many strains have been domesticated, resulting in
changes in gene expressions and colony morphology (Kuthan et al. 2003).
The dominance of cereals in traditional diets makes the energy balances
of grain production the most revealing indicators of agricultural produc-
tivity. Data on typical agricultural labor requirements and their energy
costs are available for a large variety of individual field and farmyard tasks
(box 3.2).
But this level of detail is not necessary for calculating approximate
energy balances. Using a representative average of typical net energy costs
in traditional farming works quite well. The typical energy needs of moder-
ate activities are about 4.5 times the basal metabolic rate for men, and five
times for women, or 1 and 1.35 MJ/h (FAO 2004). Subtracting the respective
basic existential needs results in net labor energy costs of 670 and 940 kJ/h.
The simple mean is roughly 800 kJ/h, and I will use it as the net food energy
cost of an average hour of labor in traditional agriculture. Similarly, gross
grain output is calculated by multiplying the harvested mass by appropriate
energy equivalents (typically by 15 GJ/t for grain with less than 15% mois-
ture that can be stored).
The ratio of these two measures indicates the gross energy return, and
hence the productivity, of these critical farming tasks. Net energy returns—
after subtracting seed requirements and milling and storage losses—were
substantially lower. Farmers had to set aside a portion of every harvest for
the next year’s seeding. The combination of low yields and high seed waste
in hand broadcasting could mean that as much one-third or even one-half
of medieval grain crops had to be set aside. With increasing harvests these
shares fell gradually to less than 15%. Some grains are eaten whole, but

The oldest Mesopotamian harness (best used with strong, short-necked
animals, and later common in Spain and Latin America) was the double
head yoke, fixed either at the front or the back of the head (fig. 3.4). That
was a primitive harness: merely a long wooden beam whose throat fasten-
ings may choke the animal in heavier labor and whose traction angle is too
large. Moreover, to avoid excessive choking of one ox or cow, the animals

Traditional Farming

71

Box 3.5

Comparing harnesses and draft power

For decades many writings repeated the claim that the ancient throat-and-

girth harness was not suitable for heavy field tasks because of its excessively

high point of traction and its choking effect created by the throat strap. This

conclusion was based on actual experiments with a reconstructed harness

done in 1910 by a French officer, Richard Lefebvre des Noëttes (1856–1936),

and published in 1924 in his book, La Force Motrice à travers les Âges. These

findings were accepted not only by many classicists but also by three leading

twentieth-century historians of technical advances, Joseph Needham (1965),

Lynn White (1978), and Jean Gimpel (1997).

But they were based on a mistaken reconstruction: experiments done by

Jean Spruytte during the 1970s with a properly reconstructed dorsal yoke har-

ness (placed directly behind the withers and fastened by chest straps) did not

result in any choking, and the harness performed well when two horses were

pulling a load of nearly 1 t (Spruytte 1977). This disproved the idea that “the

classical cultures were ‘blocked’ by a defective system of harnessing animals”

(Raepsaet 2008, 581). But in his tests, Spruytte used a light nineteenth-century

carriage (much lighter than a Roman wagon), and hence, even if the differ-

ence in horse size is ignored, his tests did not fully replicate the conditions

common two millennia ago. In any case, because of the weight limit (500 kg)

put on the horse-drawn wagons by the Theodosian Code (439 CE), “It seems

certain that the Romans were aware of the distress caused in horses by pulling

heavy loads” (Gans 2004, 179).

Traditional Farming
75
Box 3.6
Energy cost, efficiency, and performance of a draft horse
A mature 500-kg horse needs about 70 MJ of digestible energy per day to main-
tain its weight (Subcommittee on Horse Nutrition 1978). If its feed is high in
grains, this may imply just 80 MJ of gross energy intake; if it is mostly less
digestible hay then it may rise to 100 MJ. Depending on the task, feed require-
ments during the working periods were 1.5–1.9 times the maintenance need.
Brody(1945) found a 500-kg Percheron working at a rate of about 500 W
expending about 10 MJ/h. With 6 hours of work and 18 hours of rest (at 3.75
MJ/h), this adds up to about 125 MJ/day.
Not surprisingly, traditional feeding recommendations concur: at the
beginning of the twentieth century American farmers were advised to feed
their working horses 4.5 kg of oats and 4.5 kg of hay a day (Bailey 1908),
which translates to about 120 M J/day. With an average power of 500 W, a
horse would do about 11 MJ of useful work during six hours, and while an
average male human would contribute less than 2 MJ, though he could not
maintain steady exertion above 80 W and managed only brief peaks above
150 W, a horse could work steadily at 500 W and have brief peak pulls in
excess of 1 kW, an effort that would require the exertions of a dozen men.

76

Chapter 3
Irrigation
The water demand of crops depends on many environmental, agronomic,
and genetic variables, but the total seasonal need is commonly about
1,000 times the mass of the harvested grain. Up to 1,500 t of water are
needed to grow 1 t of wheat, and at least 900 t must be supplied for every
tonne of rice. About 600 t will suffice for a tonne of corn, a more water-
efficient C 4 crop and the staple grain with the highest water use efficiency
(Doorenbos et al. 1979; Bos 2009). This means that for wheat yields of
between 1 and 2 t/ha, the total water needs during the four months of the
growing season were 15–30 cm. In contrast, annual precipitation in the
arid and semiarid regions of the Middle East ranged from a mere trace to
no more than 25 cm.
Cropping in such locations thus required irrigation as soon as the fields
were established beyond the reach of seasonal floods, which saturated val-
ley soils and allowed for the maturation of one harvest or as soon as the
growing population required the planting of a second crop during the low-
water season. Irrigation was also necessary to cope with seasonal water
shortages. These are especially pronounced in the most northerly reaches of
monsoonal Asia, in Punjab, and on the North China Plain. And, of course,
rice growing required its own arrangements for flooding and draining the
fields.
Gravity-fed irrigation—using canals, ponds, tanks, or dams—requires no
water lifting and has the lowest energy cost. But in river valleys with mini-
mal stream gradients and on large cultivated plains, it has always been
necessary to lift large volumes of surface or underground water. Many lifts
were only across a low embankment, but often they had to surmount steep
stream banks or come from deep wells. Unavoidable inefficiencies, aggra-
vated by the rough finishing of moving parts and often by the absence
of lubricants, prolonged the task. Irrigation powered by human muscles
represented a large labor burden even in societies where tedious work was
the norm. Much ingenuity went into designing mechanical devices using
animal power or water flow to ease that work—as well as to make higher
lifts possible.
An impressive variety of mechanical devices was invented to lift
water for irrigation (Ewbank 1870; Molenaar 1956; Oleson 1984, 2008;
Mays 2010). The simplest ones—tightly woven or lined shovel-like scoops,
baskets, or buckets—were used to raise water less than 1 m. A scoop or a

 Fertilization
Atmospheric CO 2 and water supply carbon and hydrogen, the two elements
forming the bulk of plant tissues as new carbohydrates. But other elements
are absolutely necessary for photosynthesis. Depending on the amount
needed, they are classified as macronutrients or micronutrients. The latter
are more numerous, including above all iron, copper, sulfur, silica and
calcium. There are only three macronutrients: nitrogen, phosphorus, and
potassium (N, P, and K). Nitrogen is by far the most important one: it
is present in all enzymes and proteins and is the element to be most likely
in short supply in continuously cultivated soils (Smil 2001; Barker and
Pilbeam 2007). Harvest of 1 t/ha of wheat (a typical French or U.S. harvest
around 1800) removes (in grain and straw) about 1 kg each of calcium and
magnesium (Ca and Mg), 2.5 kg of sulfur (S), 4 kg of potassium, 4.5 kg of
phosphorus, and 20 kg of nitrogen (Laloux et al. 1980).
Rain, dust, weathering, and the recycling of crop residues would, in most
cases, replenish withdrawals of phosphorus, potassium, and micronutri-
ents, but continued cropping without fertilization would create nitrogen
deficits, and because nitrogen’s availability largely determines the grain size
and protein content, these shortages caused stunted growth, poor yields,
and poor nutritional quality. Traditional farming could replace nitrogen in
only three ways: by directly recycling unwanted crop residues, that is, by
plowing in part of straws or stalks that were not removed from the fields for

 Traditional Farming
83
feed, fuel, and other household uses; by applying a variety of organic mat-
ter, above all by spreading (usually composted) animal and human urine
and feces and other organic wastes; and by cultivating leguminous crops to
enhance nitrogen soil content for a subsequent nonleguminous crop (Smil
2001; Berklian 2008).

Crop Diversity
Modern farming is dominated by monocultures, annual plantings of the
same crop, reflecting the regional specialization of commercial agriculture….
Long experience taught many ancient cultivators the perils of monocul-
tures. In contrast, the rotation of cereals and leguminous crops either
replenishes the soil nitrogen or at least eases the drain on its soil reserves.
The cultivation of a variety of grain, tuber, oil, and fiber crops lowers the
risk of total harvest failure, discourages the establishment of persistent
pests, reduces erosion, and maintains better soil properties (Lowrance et al.
1984; USDA 2014). Crop rotations can be chosen to fit climatic and soil
conditions and to satisfy specific dietary preferences; they are highly desir-
able from an agronomic point of view, but where more than one crop is
grown per year (multicropping) they obviously require more labor. In
places with dry spells, irrigation will be needed, and for intensive multi-
cropping, with three or even four different species grown every year in the
same field, substantial fertilization will be also needed. Where two or more
crops are grown in the same field at the same time (intercropping), labor
demands may be even higher. The fundamental reward of multicropping
is its ability to support larger populations from the same amount of
cultivated land.

The twentieth-century growth of global electricity output was even
faster than the expansion of fossil fuel extraction, whose annual average
was about 3% (fig. 6.5). Less than 2% of all fuel was converted to electricity
in 1900; the share was still less than 10% in 1945, but by the century’s end
it had risen to about 25%. In addition, new hydroelectric stations (on a
large scale after World War I) and new nuclear capacities (since 1956) fur-
ther expanded electricity generation. As a result, the global electricity sup-
ply went up by about 11% a year between 1900 and 1935, and by more
than 9% annually thereafter until the early 1970s. For the remainder of the
century the growth of electricity generation declined to about 3.5% annu-
ally, largely because of lower demand in high-income economies and higher
conversion efficiencies. New ways of electricity generation from renewable
sources such as solar energy and wind have shown notable advances only
since the late 1980s.

Box 6.2
Energy costs of nitrogenous fertilizers
The energy requirements of the Haber-Bosch synthesis include fuels and elec-
tricity used in the process and energy embodied in the feedstocks. The coke-
based Haber-Bosch synthesis process in BASF’s first commercial plant required
more than 100 GJ/t NH 3 in 1913; before World War II the rate was down to
around 85 GJ/t NH 3 . After 1950 natural gas–based processes lowered the over-
all energy cost to 50–55 GJ/t NH 3 ; centrifugal compressors and high-pressure
reforming of steam and better catalysts lowered the requirements first to less
than 40 GJ/t by the 1970s, and then to around 30 GJ/t by the year 2000, when
the best plants needed only about 27 GJ/t NH 3 , close to the stoichiometric
energy requirement (20.8 GJ/t) for ammonia synthesis (Kongshaug 1998; Smil
2001). Typical new natural gas–based plants use 30 GJ/t NH 3 , about 20% more
when using heavy fuel oil, and up to about 48 GJ/t NH 3 for coal-based synthe-
sis (Rafiqul et al. 2005; Noelker and Ruether 2011).
The average performance was about 35 GJ/t in 2015; the last rate corre-
sponds to about 43 GJ/t N. But most farmers do not apply ammonia (a gas
under normal pressure) and prefer liquids or solids, especially urea, which has
the highest share of nitrogen (45%) among solid compounds that are easily
applied even to small fields. Converting ammonia to urea, packaging, and
transportation bring the overall energy cost to 55 GJ/t N. Using this rate as the
global average means that in 2015, with about 115 Mt N used in agriculture,
the synthesis of nitrogenous fertilizers claimed about 6.3 EJ of energy, or just
over 1% of the global energy supply (Smil 2014b).

At the beginning of the twenty-first century, the global harvest provided
a daily supply averaging (for a population nearly four times as large as in
1900) about 2,800 kcal/capita, more than adequate if it were equitably
accessible (Smil 2008a). The roughly 12% of the world’s population that is
still undernourished does not have enough to eat because of limited access
to food, not because of its unavailability, and the food supply in affluent
countries is now about 75% higher than the actual need, resulting in enor-
mous food waste (30–40% of all food at the retail level) and high rates of
overweight and obesity (Smil 2013a). Moreover, plenty of grain (50–60%
in affluent countries) is fed to domestic animals. Chickens are the most
efficient converters of feed (about three units of concentrate feed for a unit
of meat); the feed:meat ratio for pork is about 9, and the production of
grain-fed beef is most demanding, requiring up to 25 units of feed for a unit
of meat.
This inferior ratio is also the function of the meat:live weight ratio: for
chicken it is as high as 0.65 and for pork it is 0.53, but for large beef animals
it is as low as 0.38 (Smil 2013d). But the energy loss in conversion to meat
(and milk) has its nutritional rewards: the rising consumption of animal
foods has brought high-protein diets to all rich nations (evident in taller
statures) and has assured, on average, adequate nutrition even in most of
the world’s largest poor populous countries. Most notably, the energy con-
tent of China’s average per capita diet is now, at about 3,000 kcal/day, about
10% ahead of the Japanese mean (FAO 2015a).

Economic Growth
To talk about energy and the economy is a tautology: every economic activ-
ity is fundamentally nothing but a conversion of one kind of energy to
another, and monies are just a convenient (and often rather unrepresenta-
tive) proxy for valuing the energy flows. Not surprisingly, Frederick Soddy,
a Nobelian physicist approaching the discipline from this perspective,
argued that “the flow of energy should be the primary concern of
economics” (Soddy 1933, 56). At the same time, energy flow is a poor mea-
sure of intellectual activity: education certainly embodies a great deal of
energy expended on its infrastructures and employees, but brilliant ideas
(which are by no means directly related to the intensity of schooling) do
not require large increases of the brain’s metabolic rate.
This obvious fact explains much of the recent decoupling of GDP growth
from overall energy demand: we impute much higher monetary values to
the nonphysical endeavors that now constitute the largest share of the eco-
nomic product. In any case, energy has been of marginal concern in mod-
ern economic studies; only ecological economists have seen it as their
primary focus (Ayres, Ayres, and Warr 2003; Stern 2010). And the public
concern about energy and the economy has been disproportionately
focused on prices in general, and on the prices of crude oil, the world’s most
important traded commodity, in particular.
In the West it was OPEC’s two rounds of oil price increases during the
1970s—both the source of Middle Eastern consumption excesses and a
threat to the region’s stability—that became a particular object of critique,
blamed for economic dislocations and social turmoil. But OPEC’s price rise
had a salutary (and long overdue) effect on the efficiency with which the
countries importing OPEC oil consumed refined fuels. In 1973, after four
decades of slow deterioration, the average specific fuel consumption of new
American passenger cars was higher than in the early 1930s, 17.7 L/100 km
versus 14.8 L/100 km, or, in American usage, 13.4 mpg versus 16 mpg (Smil
2006)—a rare example of a modern energy conversion becoming less
efficient.
Higher oil prices forced the reversal, and between 1973 and 1987 the
average fuel demand of new cars on the North American market was cut in
half as the CAFE (Corporate Automobile Fuel Efficiency) standard fell to 8.6
L/100 km (27.5 mpg). Unfortunately, the post-1985 fall in oil prices first
stopped and then even reversed (with more SUVs and pickups) this effi-
ciency progress, and return to rationality came only in 2005. And OPEC’s
price rise had a beneficial effect for the global economy as it significantly
reduced its average oil intensity (amount of oil used per unit of GDP). Power
plants stopped burning liquid fuels; iron makers replaced injections of fuel
oil to blast furnaces by powdered coal; jet engines became more efficient;
and many industrial processes converted to natural gas. The results have
been quite impressive. By 1985 the U.S. economy needed 37% less oil to
produce a dollar of GDP than it did in 1970; by the year 2000 its oil inten-
sity was 53% lower; and by 2014 it required 62% less crude oil to create a
dollar of GDP than it did in 1970 (Smil 2015c).

And (a curiously neglected fact) Western governments have been mak-

ing more money from oil than OPEC. In 2014, taxes in G7 countries

accounted for about 47% of the price of a liter of oil, compared to about

39% going to the producers, with the respective national shares at 60/30 in

the UK, 52/34 in Germany, and 15/61 in the United States (OPEC 2015).

Moreover, to ensure a secure supply, many governments (including those of

market economies) have engaged in a great deal of industry regulation,

while governments in many oil-producing countries have been buying

political support with heavy subsidies of energy prices (GSI 2015). Saudi

subsidies claimed more than 20% of all government expenditures in 2010,

and China’s coal subsidies have resulted in prices fixed even below the

production cost.

Growth—its origins, rate, and persistence—has been the leading concern

of modern economic inquiries (Kuznets 1971; Rostow 1971; Barro 1997;

Galor 2005), and hence the links between energy consumption and the

increase gross economic product (either gross domestic product, GDP, for

individual economies, or GWP, gross world product, for studying global

trends) have received a great deal of attention (Stern 2004, 2010; World

Economic Forum 2012; Ayres 2014). Traditional preindustrial economies

were either largely stationary or managed to grow by a few percent per

decade, and average per capita energy consumption advanced at an even

slower pace: there is no shortage of testimonies from the early decades of

the nineteenth century showing that the living conditions of some impov-

erished groups were not very different from those that had prevailed even

two or three or four centuries before.

In contrast, the fossil-fueled economies have seen unprecedented rates

of growth, though modified by the cyclical nature of economic expansion

(van Duijn 1983; ECRI 2015) and interrupted by major internal or interna-

tional conflicts. Industrializing societies of the nineteenth century saw

their economies growing by 20–60% in a decade. Such growth rates meant

that the output of the British economy in 1900 was nearly ten times larger

than in 1800. America’s GDP doubled in just 20 years, between 1880 and

1900. Japanese output during the Meiji era (1868–1912) rose 2.5 times. Eco-

nomic growth during the first half of the twentieth century was affected by

two world wars and the great economic crisis of the 1930s, but there had

never been a period of such rapid and widespread growth of output and

prosperity as between 1950 and 1973.

The steady pre-1970 decline in real crude oil prices was a critical ingredi-

ent of this unprecedented expansion. American per capita GDP, already the

world’s highest, rose by 60%. The West German rate more than tripled, and

 Ayres, Ayres, and Warr (2003) identified the declining price of useful
work as the growth engine of the U.S. economy during the twentieth cen-
tury, useful work being the product of exergy (the maximum work possible
in an ideal energy conversion process) and conversion efficiency. Once the
historical data of economic output are normalized (with GDP values
expressed in constant, inflation-adjusted monies and with the national
products used to calculate GWP given in terms of purchasing power parity
rather than by using official exchange rates), impressively strong long-term
correlations between economic growth and energy use emerge on both
global and national levels.
Between 1900 and 2000 the use of all primary energy (after subtracting
processing losses and nonfuel uses of fossil fuels) rose nearly eightfold, from
44 to 382 EJ, and the GWP increased more than 18 times, from about $2
trillion to nearly $37 trillion in constant 1990 monies (Smil 2010a; Maddi-
son Project 2013), implying an elasticity of less than 0.5. High correlations
of the two variables can be found for a single country over time, but the
elasticities differ: during the twentieth century, the Japanese GDP increased
52-fold and total energy use rose 50-fold (an elasticity very close to 1.0),
while the multiples for the United States were, respectively, nearly 10-fold
and 25-fold (an elasticity of less than 0.4), and for China nearly 13-fold and
20-fold (an elasticity of 0.6

Accounting for growth: the role of physical work
Robert U. Ayres∗, Benjamin Warr1
Center for the Management of Environmental Resources, INSEAD, Boulevard de Constance,
77305 Fontainebleau, France
Received 1 September 2002; received in revised form 1 August 2003; accepted 1 October 2003
Available online 22 February 2004
Abstract
This paper tests several related hypothesis for explaining US economic growth since 1900. It begins
from the belief that consumption of natural resources—especially energy (or, more precisely,
exergy) has been, and still is, an important factor of production and driver of economic growth.
However the major result of the paper is that it is not ‘raw’ energy (exergy) as an input, but exergy
converted to useful (physical) work that—along with capital and (human) labor—really explains
output and drives long-term economic growth. We develop a formal model (Resource-EXergy Service
or REXS) based on these ideas. Using this model we demonstrate first that, if raw energy
inputs are included with capital and labor in a Cobb–Douglas or any other production function satisfying
the Euler (constant returns) condition, the 100-year growth history of the US cannot be
explained without introducing an exogenous ‘technical progress’ multiplier (the Solow residual) to
explain most of the growth. However, if we replace raw energy as an input by ‘useful work’ (the
sum total of all types of physical work by animals, prime movers and heat transfer systems) as a
factor of production, the historical growth path of the US is reproduced with high accuracy from
1900 until the mid 1 970s, without any residual except during brief periods of economic dislocation,
and with fairly high accuracy since then. (There are indications that an additional factor,
possibly information technology, needs to be taken into account as a fourth input factor since the
1970s.) Various hypotheses for explaining the latest period are discussed briefly, along with future
implications.
© 2004 Elsevier B.V. All rights reserved.
JEL classification: 011; 013; 014
Keywords: Exergy; Technology; Economy; Growth; Efficiency; Dematerialisation

The most important lesson to be drawn from looking at long-term trends
of per capita energy use and economic growth is that respectable rates of
the latter can be achieved with progressively lower use of the former. In the
United States a continuing if slow population growth has brought further
increases in the absolute consumption of fuels and electricity, but the aver-
age per capita use of primary energy has been flat (with only minor fluctua-
tions) for three decades, since the mid-1980s, yet the real GDP (in chained
2009 dollars) per capita gained nearly 57%, growing from $32,218 in 1985
to $50,456 in 2014 (FRED 2015). Similarly, in both France and Japan (where
population is now declining) per capita primary energy use has stabilized
since the mid-1990s—yet in the following two decades the average per cap-
ita GDP increased, respectively, by about 20% and 10%.
But these outcomes must be interpreted with caution as those periods
of relative energy-GDP decoupling coincided with extensive offshoring of
U.S., European, and Japanese energy-intensive heavy industries and manu-
facturing to Asia in general and to China in particular: it would be prema-
ture to conclude that the recent experience of those three major economies
is a harbinger of a widespread decoupling trend. And mainly because of
China’s enormous growth in pre-2014 energy demand (achieving a nearly
4.5-fold increase since 1990), the global primary energy supply had to rise
nearly 60% in order to produce a 2.8-fold rise in GWP during the 25 years
after 1990 (an elasticity of 0.56). Moreover, declines in electricity intensity
have been much slower than the declines in overall energy intensity.
Between 1990 and 2015 the global drop was just short of 20% (compared to
>40% for all energy), and the U.S. decline was also 20%, but rapidly mod-
ernizing China saw no decline between 1990 and 2015.

Energy flows limit but do not determine the biospheric organization on
any scale. As Brooks and Wiley (1986, 37–38) put it,
Energy flows do not provide an explanation for why there are organisms, why or-
ganisms vary, or why there are different species. … It is an organism’s intrinsic
properties that determine how energy will flow, not the opposite. If the flow of en-
ergy were deterministic for biological systems, it would be impossible for anything
living to starve to death. … We suggest that organisms are physical systems with
genetically and epigenetically determined individual characteristics, which utilize
energy that is flowing through the environment in a relatively stochastic manner.

Georgescu-Roegen (1980, 264) suggested a fine analogy that also cap-
tures the challenge of historical explanations: geometry constrains the size
of the diagonals in a square, not its color, and “how a square happens to be
‘green’, for instance, is a different and almost impossible question.” And so
every society’s field of physical action and achievement is bound by the
imperatives arising from the reliance on particular energy flows and prime
movers—but even small fields can offer brilliant tapestries whose creation
is not easy to explain. Mustering historical proofs for this conclusion is
easy, in matters both grand and small.

Feeding the World. Malthus wasn´t such a miserable old misanthrope apparently.Vaclav Smil, Albert Bartlett Energy Economics and no Alarmism.

123