estimates made by Stolp and Eppenga 1998 on productivity of natural forests in different parts of
the world. The data used for built-up land refer only to domestic land use; time trends were only
available for the N etherlands. F or the other three countries, domestic land use has been assumed to
grow along with the size of the population. The exact calculations, assumptions and data sources
are described in van Vuuren et al. 1999a.
3
.
2
. EF for carbon diox ide emissions F or the 1985 – 1994 period for the N etherlands,
earlier calculations
by R IVM
and Statistics
N etherlands R IVM , 1998; CBS, 1998 could be used in which the EF for carbon dioxide emis-
sions was assessed on the basis of: 1 consump- tion statistics, 2 energy intensities per product
and 3 types of energy consumed. F or 1980, we have estimated net carbon dioxide emissions on
the basis of normal carbon dioxide emissions statistics and trends in the 1985 – 1997 period.
D omestic carbon dioxide emissions in Benin, Bhutan and Costa R ica have been calculated us-
ing energy consumption statistics as collected by van Vuuren and de K ruijf 1998. F or carbon
dioxide emissions related to net imports of manu- factured goods in Costa R ica, we have used the
results of Wackernagel et al. 1997. F or Bhutan and Benin, we concluded on the basis of available
trade statistics that the carbon dioxide emissions attributed to the net import of products will be
relatively small. H owever, we considered available statistics and energy intensities which have been
determined for industrialised countries only to be insufficiently reliable to calculate the size of these
emissions
— and
thus only used domestic
emissions.
4. Results
4
.
1
. EF for real land use On a global scale, available land per person is
decreasing rapidly due to population growth, changes in production and consumption patterns,
expansion of settlements and land degradation U N EP, 1997. Earlier assessments of land-use in
the N etherlands H arjono et al., 1996; van den H oek et al., 1996; Wackernagel et al., 1997 show
that the most important type of landuse is land required for food production and wood products.
H arjono et al. 1996 indicate that landuse for minerals and direct landuse for tourism for the
N etherlands can be ignored. F ig. 2 shows the total amount of land used for consumption of food,
animal and wood products and built-up land in Benin, Bhutan, Costa R ica and the N etherlands
for 1980, 1987 and 1994 on the basis of the data as described in Section 3.
F ig. 2 shows that the net EF for land use increases for all countries. In all of them, the
largest shares of land use for those included in the study can be attributed to the consumption
of wood 25 – 50 of land use and animal prod- ucts 20 – 55 of land use. It should be noted that
products can be used in a totally different way. In Benin and Bhutan, for example, land for wood
consumption is almost completely caused by con- sumption of fuel wood — while in the N ether-
lands almost all wood is used as industrial round wood in particular for paper production. In
Bhutan and Costa R ica, use of pasture land for low-intensity
husbandry explains
the large
amount of land attributed to these categories. Cereals and other crops account for about 10 –
30 of land use. The land attributed to built-up land is very low 1 – 3.
As discussed in Section 2, comparison of the national footprints to the area of land actually
available reveals to what extent the country relies on resources outside its own territory. F or con-
sumption in the N etherlands more land is used than domestically available; this was to be ex-
pected in view of the high population density. By means of international trade, the N etherlands has
historically been able to use additional land in other countries — for instance, by importing
cassava as fodder. According to an earlier assess- ment similar methodology and results as in this
assessment, focusing more on national data the total amount of land used in 1995 for the N ether-
lands’ economy for food, animal feed and wood production inside and outside its borders gross
EF was equal to three to four times its land
F ig. 2. N et land use for agricultural products and wood. H orizontal line indicates total surface area of each country.
surface Booy and R os, 1997. By subtracting the land use for the N etherlands’ exports to other
countries, the net EF for land use is found to be about two to three times the total land surface of
the N etherlands. The largest share of land outside the N etherlands used for D utch consumption is
land for timber from Scandinavia and other parts of Western Europe. F ood production also claims
an important share, in particular, in N orth and South America, and Africa. F inally, land is used
for the production of animal feed in Western Europe, N orth America, South America and
Southeast Asia Thailand.
The surplus of domestically available land and land used for human consumption in Benin,
Bhutan and Costa R ica is mainly brought about by relatively large areas that are still undomesti-
cated. Also for Costa R ica, a sizeable amount of land is allocated to exports of fruit and coffee. At
first glance, the surplus of land in all three coun- tries might indicate that they still have the ability
to increase the amount of land used within their own territory although at the expense of biodi-
versity. H owever, in Bhutan, in particular, the potential for agricultural expansion is severely
limited due the lack of arable land.
4
.
2
. Producti6ity per hectare Agricultural productivity is partly a function of
human influenced factors, such as technology, improved crops and fertilisers. F or the countries
included in this study, productivity of most prod- ucts is highest in the countries with the highest
level of consumption, the N etherlands and Costa R ica. Over the study period, productivity in-
creased in all four countries as a result of various factors, including a heavy increase in the use of
fertilisers in Costa R ica and Benin van Vuuren and de K ruijf, 1998. In addition to human influ-
enced factors, also natural circumstances can strongly influence productivity for instance, flat,
highly productive land in the N etherlands versus savannah in Benin or alpine pastures in Bhutan.
It might be possible to calculate a ‘corrected’ land use using ‘natural yield factors’, but this would
require a large set of assumptions. H ere, we have simply added the land use based on global aver-
age yields as a reference — thus not only nullify- ing differences in natural circumstances, but also
management factors also Wackernagel et al., 1997 calculate their figures on the basis of global
yields. The combination of real yield and global
F ig. 3. N et land use per capita based on actual global average yield 1994. H orizontal line indicates global average land use per capita.
average yields figures left and right parts of F ig. 3 indicates how important differences in yield
are: for the N etherlands, for instance, land use increases from 0.7 to almost 3.0 ha per capita.
Possibly surprisingly, F ig. 3 shows that land use per capita is lowest for the N etherlands as a result
of the high agricultural productivity of this coun- try. F or Costa R ica, the land use per capita is
highest, mainly due to extensively used pasture lands which are often marginal lands cleared
from forest over the last decades. F ig. 3 also shows that in all four countries, the amount of
land used per capita declines.
In terms of net land use based on global aver- age yields, obviously land use of the N etherlands
is largest — indicating the high level of consump- tion in this country. In particular, the consump-
tion of meat and diary products represents a large share of the land use of the N etherlands if ac-
counted for on the basis of global yields. Both the N etherlands and Costa R ica use a larger amount
of ‘global average yield land’ than is actually available. On this measure, land use of Benin and
Bhutan are lowest, being directly related to the low consumption level in these countries.
4
.
3
. Uncertainty in the estimated land use We calculated the highest and lowest estimate
for land use in each country, which might give some idea of the uncertainty involved. F or the
N etherlands, these respective estimates are about 20 lower and 50 higher than the central esti-
mate — mainly as a result of uncertainty in the estimates on the productivity of forests and the
allocation of land to exports of dairy products and meat. F or Benin and Costa R ica, the high
and low estimates differ by about 30 – 50 from the central estimate — for Bhutan this is about
40 – 60. Again, the productivity attributed to forests dominates the differences found.
4
.
4
. T rends in the
1980
–
1994
period agricultural
products F our trends could be responsible for changes in
the total real net land use for agricultural prod- ucts: changes in the total population of each of
the countries, changes in yields, changes in the type of products consumed and changes in the
total amount of products consumed.
Table 2 shows the decomposition of changes into these factors, but, again, limited by the reli-
ability of the data. The figures should be seen as rough indications. In all four countries, increasing
population alone has caused an important up- ward pressure on land use about 10 increase in
the N etherlands and 40 – 50 increase in the other three countries. At the same time, land use per
capita has decreased in all four countries by 10 –
Table 2 Trends in net land use driving forces 1980–1994
a
Benin N etherlands
Costa R ica Bhutan
change change
change change
−1 Consumption per capita
15 5
−10 Changes in consumption pattern
−1 5
−30 −5
Increase in productivity −25
−9 −11
Land use per capita −15
−12 −15
Population 40
50 45
9 25
−3 Land use
25 20
a
The effect of consumption per capita, changing consumption patterns and productivity increase have been estimated by keeping the other two factors constant at 1987 levels and thus analysing the influence of each factor separately.
17. Increasing yield per product has been the most important factor behind the decrease in per
capita land use in Benin, Costa R ica and the N etherlands — while in Bhutan, available data
suggest a further shift away from land-intensive dairy products, also contributing to a lower land
use per capita.
4
.
5
. Ecological footprint of carbon diox ide emissions
N ot surprisingly, F ig. 4 shows that far more carbon dioxide is emitted for consumption by a
person in the N etherlands than a person in Costa R ica 60 less and Benin and Bhutan more than
90 less. This is not only a result of lower energy consumption, but also because energy consump-
tion is based less on fossil fuels. M ost of the commercial energy consumed in these countries is
based on hydropower thus zero emissions.
9
In addition, a lot of fuelwood is consumed in
households.
10
The carbon dioxide emissions attributed to con- sumption in the N etherlands are considerably less
than the emissions within this country. This is caused by the N etherlands’ relatively large heavy
industry, which produces goods for many Eu- ropean countries. F or Costa R ica net carbon
dioxide emissions are slightly higher than domes- tic emissions as a result of import of manufac-
tured products. As for land use, the level of carbon dioxide emissions is also influenced by
differences in natural circumstances a cold cli- mate will increase heating requirements, although
these influences are of limited importance.
F ig. 4. EF for carbon dioxide emissions in 1994. H orizontal line indicates global average per capita emissions.
10
Carbon dioxide emissions from fuelwood are often not included in carbon dioxide emission inventories. In Benin and
Bhutan, energy consumption of fuelwood is about 10 G J per person per year in comparison with the 180 G J per person of
commercial fuels consumed in the N etherlands. Land use needed for fuelwood consumption has been included in the
forest areas in F igures 3 and 4.
9
The use of land for hydropower has not been included in this study. Wackernagel et al. 1997 estimate the global
average area occupied by hydropower dams is equal to 0.001 haG J per year. U sing this world average figure for each of the
four countries suggests per capita use of land for hydropower as negligable.
Table 3 Trends in carbon dioxide EF driving forces 1980–1994
a
Benin Bhutan
N etherlands Costa R ica
change change
change change
52 Population growth capita
40 34
9 G D P per capita ppp 1987capita
76 −15
2 20
– 31
−10 −9
Carbon intensity M ton CO
2
ppp 1987 Carbon dioxide emissions M ton
15 –
80 18
a
Sources: carbon dioxide emissions are calculated as described under methodology; population growth and G D P in purchasing- power parity dollars per capita are taken from World Bank 1998 data for Bhutan from van Vuuren and de K ruijf, 1998.
4
.
6
. Focusing on the N etherlands The data for the N etherlands were also available
for the 1985 – 1997 period domestic emissions and net emissions for consumption of the population.
Earlier, Suri and Chapman 1998 reported that one factor behind the decreasing elasticity of energy
to G D P in industrialised countries, as a group, is an increased import of manufactured goods from
developing countries. This would imply — in terms of the EF — that net emissions grow faster than
domestic emissions since the former would also include the increased emissions in developing coun-
tries. The available data for the N etherlands, however, does not provide evidence for such a shift
for this particular country. The N etherlands, how- ever, has a large positive balance of trade; increas-
ing exports in the same period may have offset increasing imports from developing countries.
4
.
7
. T rends in the
1980
–
1994
period Trends in carbon dioxide emissions can be dis-
cussed vis-a`-vis changes in population size popula- tion, per capita income G D Ppopulation and
carbon intensity CO
2
emissionsG D P Eq. 2, using the so-called K aya identity; K aya, 1989.
CO
2
emissions = population × G D P
population ×
CO
2
emissions G D P
2 Although this identity does not provide an explana-
tory model, it offers a simple and transparent accounting framework, which can assess the contri-
bution of different factors to the overall change in carbon dioxide emissions.
Table 3 shows the changes of the three different factors. In all four countries, population growth
alone all other factors considered to be constant can be identified as an important upward pressure
on carbon dioxide emissions. At the same time, the other factors change also as a result of which, for
instance, Benin, the county with the highest popu- lation growth, has only a relatively small increase
in carbon dioxide emissions. One of them is the change in activities — measured in terms of G D P.
In the N etherlands and Bhutan G D P per capita has increased strongly over the 1980 – 1994 period less
so in Benin and Costa R ica as a result of strong economic recessions during the period of study.
The remaining factor, carbon intensity carbon dioxide emissions per unit of G D P, is a function
of energy efficiency, structural changes in the economy and the type of fuel used. World-wide,
carbon intensities decrease over time for industri- alised countries while for developing countries
trends are country-specific. Carbon intensity has decreased in the N etherlands and Benin, but in-
creased in Costa R ica as result of an increased share of commercial fuels. The latter can be explained by
the transition in Costa R ica towards an industri- alised economy.
4
.
8
. A ggregation into W ackernagel and R ees
’
eco- logical footprint
F ig. 5 shows the EF that would result from our calculations based on the definition of Wacker
F ig. 5. Ecological footprint according to the definition Wackernagel R ees. H orizontal line indicates sustainable global level according to Wackernagel and R ees 1996.
F ig. 6. Wackernagel and R ees’ EF on the basis of this study and earlier results.
nagel and R ees, thus including hypothetical land for carbon sequestration ‘energy land’. ‘Energy
land’ now becomes an important part of the EF of the N etherlands — and to some degree as well the
EF of Costa R ica. The calculated figures can be compared with the 1.31 ha per capita considered
sustainable at global level by Wackernagel et al. 1997.
11
The N etherlands and also Costa R ica are considerably above this level on a global average
yield basis. The land use based on global average yields as
calculated in this study has been compared to the earlier calculations of Wackernagel et al. 1997.
12
The aggregated EF s of this study for
the
11
Wackernagel et al. 1997 mention that on global scale, 2.1 ha. per capita was available in 1993 — including 0.56 ha
sea. After subtracting the sea area, setting aside 12 of the land area for biodiversity, and recalculating available land on
the basis of 1994 population data, 1.31 ha per capita remains.
12
F or comparison, we included the equivalence factors of Wackernagel et al. 1997 in our results. M oreover, we sub-
tracted the use of sea area for fish from the results of Wacker- nagel et al. 1997.
N etherlands and Costa R ica are somewhat lower than the earlier results 4.6 ha per capita versus
5.2 ha per capita, and 2.1 ha per capita versus 2.4 ha per capita, respectively. The main under-
lying differences are a smaller area attributed to forest use and carbon dioxide emissions for both
countries and a larger area attributed to pasture land for Costa R ica. F ig. 6 shows the results of
this study in comparison with the 52 countries included in the study of Wackernagel et al.
1997. In general, countries with a higher G D P per capita score higher, both with respect to the
carbon dioxide emissions and the land use com- ponent of the EF . At the same time, however,
the spread of individual countries is large. The discussion in this article gives some idea of the
causes of the spread, including consumption pat- terns and the overall level of consumption, local
agricultural and climatic agricultural practices, influence of environmental policies and uncer-
tainty in data and calculation methods. In case of land use, the available data seem to suggest a
slow saturation at higher income levels. Within the N etherlands, a study looking at land use of
different income groups made a similar observa- tion for land use for food production; land use
for wood products and paper was simply corre- lated with income R IVM , 1998. The countries
above the ‘average curve’ are all large, scarcely populated countries — perhaps caused by the
fact that land efficiency in these countries is less of a need. The carbon dioxide component shows
that in addition to countries with high domestic emissions, such as the U SA and the R ussian
F ederation, also affluent, importing countries can, in EF terms, be identified as having a large
per capita contribution to global emissions. This includes, for instance, countries like H ong-K ong,
Iceland and Singapore.
5. Discussion and conclusions
