Ecological Economics 101 (2014) 4353

Contents lists available at ScienceDirect

Ecological Economics
journal homepage: www.elsevier.com/locate/ecolecon

Analysis

Valuing the impact of trade on local blue water

Anne Biewald a,, Susanne Rolinski a, Hermann Lotze-Campen a,b, Christoph Schmitz a, Jan Philipp Dietrich a
a
Potsdam Institute for Climate Impact Research, Telegraphenberg A 31, 14473 Potsdam, Germany
b
Humboldt University of Berlin, Unter den Linden 6, 10099 Berlin, Germany

a r t i c l e i n f o a b s t r a c t

Article history: International trade of agricultural goods impacts local water scarcity. By quantifying the effect of trade on crop
Received 7 August 2012 production on grid-cell level and combining it with cell- and crop-specic virtual water contents, we are able
Received in revised form 6 January 2014 to determine green and blue water consumption and savings. Connecting the information on trade-related
Accepted 6 February 2014
blue water usage to water shadow prices gives us the possibility to value the impact of international food crop
Available online 14 March 2014
trade on local blue water resources. To determine the trade-related value of the blue water usage, we employ
Keywords:
two models: rst, an economic land- and water-use model, simulating agricultural trade, production and
Virtual water water-shadow prices and second, a global vegetation and agricultural model, modeling the blue and green virtual
Blue and green water water content of the traded crops. Our study found that globally, the international trade of food crops saves blue
Water scarcity water worth 2.4 billion US$. This net saving occurs despite the fact that Europe exports virtual blue water in food
Agricultural trade crops worth 3.1 billion US$. Countries in the Middle East and South Asia prot from trade by importing water in-
tensive crops, countries in Southern Europe on the other hand export water intensive agricultural goods from
water scarce sites, deteriorating local water scarcity.
2014 Elsevier B.V. All rights reserved.

1. Introduction
Water scarcity is a local phenomenon susceptible to global food pro-
duction and its changes, since agriculture has the largest share in the
consumption of global freshwater resources (Molden, 2007). With
a growing world population and changes in dietary habits (Pingali,
2007), the demand for agricultural production and thus the demand
for fresh water will increase in the future (Rosegrant and Sombilla,
1997; Vrsmarty et al., 2000). In this context international trade of
virtual water, i.e. water embedded in agricultural goods, as dened by
Allan (1996), plays an important role for local water availability now
and in the coming decades. According to Hoekstra and Mekonnen
(2012), roughly one quarter of the water used in global agricultural pro-
duction can be assigned to virtual water exports.
Falkenmark and Lannerstad (2010) estimate that it will be necessary
by 2050 to double the virtual water trade in order to compensate
agricultural water decits. In this sense the International Water
Management Institute (IWMI) and the Government Ofce for Science
(Government Ofce for Science, London, 2011; Molden, 2007) both
state that an increase in global food trade and the consequent virtual
water ows will offer the possibility of relieving water stress and a
more efcient use of global water.
Corresponding author.
E-mail addresses: Anne.Biewald@pik-potsdam.de (A. Biewald),
Susanne.Rolinski@pik-potsdam.de (S. Rolinski), Lotze-campen@pik-potsdam.de
(H. Lotze-Campen), Christoph.Schmitz@pik-potsdam.de (C. Schmitz),
Jan.Dietrich@pik-potsdam.de (J.P. Dietrich).
As several studies show, the main share of the virtual water in agri-
cultural trade is green water (precipitation-derived soil water), while
the share of blue water (runoff-derived liquid water resources)1 is
relatively small (Aldaya et al., 2010; Fader et al., 2011; Hanasaki et al.,
2010; Hoekstra and Mekonnen, 2012; Yang et al., 2006). These results
highlight the relevance of rainfed agriculture and, therefore, of land
management in addition to blue water management.
Virtual water trade and the respective savings through trade of agri-
cultural goods are quantied in a number of studies. Global estimates of
virtual water ows related to crop trade were given e.g. by Hoekstra and
Hung (2005), Yang et al. (2006), Aldaya et al. (2010), Fader et al. (2011)
and Mekonnen and Hoekstra (2011). The respective savings related
to different agricultural goods were determined by Oki and Kanae
(2004), Chapagain and Hoekstra (2008), Hanasaki et al. (2010) as well
as Hoekstra and Mekonnen (2012).
In recent years, the distinction between green and blue water was
taken into consideration for calculating virtual water ows. Virtual
blue and green water exports were estimated for all agricultural goods
together (Hanasaki et al., 2010; Hoekstra and Mekonnen, 2012) as well
as for different crop types (Fader et al., 2011; Mekonnen and Hoekstra,
2010, 2011).
The calculation of virtual water ows in the literature cited above
is based on national trade statistics and average virtual water contents
of the export countries. The respective savings are calculated as the dif-
ference of virtual water used in the exporting countries and the virtual
1
As dened in Falkenmark and Rockstrm (2004).

http://dx.doi.org/10.1016/j.ecolecon.2014.02.003
0921-8009/ 2014 Elsevier B.V. All rights reserved.
44 A. Biewald et al. / Ecological Economics 101 (2014) 4353
water potentially used in the import countries. These approaches are
limited for different reasons. First, they use national average values for
water productivity instead of explicit cell based productivities. This be-
comes especially problematic in large countries with different climatic
zones, which relate to different water productivities. Second, calculating
virtual water savings by using bilateral trade data, it is not possible
to take into account that the landuse pattern and the water usage
would change with different production and productivities. Thus, the
calculated savings might be awed. Third, the crop consumption of live-
stock is not taken into account. When a country is forced to produce the
livestock itself, it might need a different amount of crops to feed the live-
stock, based on a specic conversion efciency. This again will inuence
the amount of water used in the crop production. For the estimation of
blue and green water savings, an additional problem accrues. These
water savings are calculated based on the simplifying assumption that
importing and exporting countries produce their crops with the same
type of water, which leads to an overestimation of the type of water
used in the export countries. Consequentially, estimating blue and
green water savings has remained an exception, only de Fraiture et al.
(2004) have tried to use this approach to estimate blue water savings
for cereals. In fact, the estimation of positive (water saved through
trade) and negative (water used for the production of export goods) vir-
tual water savings on grid cell level has not only been assessed by many
authors as an inevitable step to avoid biased water ows (Fader et al.,
2011; Oki and Kanae, 2004) but even as not available on global scale
(Islam et al., 2007). Therefore, Oki and Kanae (2004) explicitly state
the importance of estimating virtual water transfers on subnational
levels. Although some studies have quantied green and blue virtual
water ows, it has remained a challenging task to estimate blue and
green water savings.
In our study, we estimate blue and green water savings for all
agricultural crops on a grid-cell level by extracting the trade-related
production and multiplying it with the cell-specic virtual water con-
tent. With our model-based (instead of data-based) approach, taking
also into account the trade of feed and livestock, we can, for the rst
time, consistently determine green and blue water savings on grid-cell
level, taking into account the difference in the origin of water used in
importing and in exporting countries.
When water savings occur in the right places, they can have a
decisive impact on local water scarcity. Being the topic of a large body
of literature, water scarcity can be measured in different ways. In
some studies, total freshwater resources are related to per capita
requirements (Falkenmark et al., 1989) or the water withdrawal-to-
availability ratio is calculated in order to indicate water scarcity
(Alcamo et al., 2003; Hanasaki et al., 2008; Oki and Kanae, 2006;
Vrsmarty et al., 2000). Schmitz et al. (2013) developed an agro-
economic water scarcity indicator on a spatially explicit level, which is
able to quantify the pressure on global water resources. Hoekstra et al.
(2012) analyzed the water consumption of river basins, incorporating
environmental ow requirements on a monthly basis to assess global
water scarcity.
The incorporation of water stress characterisation factors is essen-
tial in linking global agricultural consumption to freshwater scarcity
(Ridoutt and Pster, 2010). Several studies have investigated the rela-
tionship between water resource availability and virtual water trade.
de Fraiture et al. (2004) and Chapagain and Hoekstra (2008) have re-
lated trade ows to national water scarcity values. Yang et al. (2003)
have linked per capita available water resources with per capita net
cereal imports. All of these studies have in common that they base
their import and export values, the water scarcity and the water
productivities on national statistics. This approach becomes problematic
as soon as countries exceed a size where water scarcity and virtual
water contents are not homogeneously distributed within these coun-
tries. Studies show that in large countries like China and in India, with
water rich as well as water scarce regions, virtual water can ow
away from water scarce areas (Ma et al., 2006; Verma et al., 2008).
Such results could not have been possible by looking at national trade
data only.
Recent literature has tried to circumvent the limitation of national
data by disaggregating data or by concentrating on specic products
or countries. Islam et al. (2007) have investigated the impact of interna-
tional trade of selected agricultural goods on water availability. The
authors disaggregated national virtual water trade data to a subna-
tional level and combined them with grid-based model results on
water availability in order to estimate the impact of trade on water
consumption. The trade-related water consumption is then combined
with the Falkenmark index for six world regions. Ridoutt and Pster
(2010) developed a stress weighted water footprint for two specic
products produced in Australia by including the water scarcity of the
different production sites. Garrido et al. (2010) estimated an economic
scarcity value of the agricultural blue virtual water exports for differ-
ent regions in Spain by multiplying literature based scarcity values
with blue virtual water exports.
In our study we go further than previous studies on virtual water
trade and water scarcity. For the rst time, we are able to determine
on a grid-based subnational level the positive and negative savings
of virtual blue and green water through international trade of crops,
livestock and feed. In addition, we estimate the impact of trade on
blue water resources by combining an economic scarcity indicator
with trade related blue water consumption. Since we combine a
biophysical water and vegetation model with an economic water and
landuse model, we can use consistent information for our analysis and
enhance therefore the reliability of our results.
2. Methods and Data
2.1. Model Description of MAgPIE
MAgPIE (Model of Agricultural Production and its Impact on the
Environment) is a global, spatially explicit, economic landuse model
optimizing in a recursive-dynamic mode (Lotze-Campen et al., 2008)
(a simplied graphical representation can be found in Fig. 1). The
model is implemented in the algebraic modeling language GAMS
(Brook et al., 1988) and the programming language R (R. Development
Core Team, 2011). The model distinguishes ten world regions on the de-
mand side (Fig. 1) and solves grid-specic (up to 0.5 degree resolution)
on the supply side.
With income and population projections (based on the ADAM pro-
ject (van Vuuren et al., 2009)) as exogenous inputs, required demand
is projected in the future and produced by 15 food crops, 5 livestock
products, ber, and fodder as intermediate input (Table 1). Feed require-
ments for the livestock production activities consist of a mixture of pas-
ture, fodder, and food crops. The livestock-specic requirements depend
on not only biological needs for maintenance and growth but also tem-
perature effects and the use of extra energy for grazing (Wirsenius,
2000). The implementation in MAgPIE is described in Weindl et al.
(2010). The model simulates time steps of 10 years and uses in each pe-
riod the optimal land-use pattern from the previous period as initial con-
dition. On the biophysical side, the model is linked to the grid-based
dynamic vegetation model LPJmL (description in Section 2.2), which
simulates crop yields depending on climatic conditions on a 0.5 degree
resolution. In addition to crop yields, LPJmL transfers water inputs, like
water availability and requirements per cell and crop, to MAgPIE.
The objective function of MAgPIE minimizes global costs, which in-
volves production costs for the agricultural commodities, technological
change costs, land expansion costs and trade and transport costs. There
are no explicit irrigation costs, but irrigation renders higher yields and
the existence of blue water is therefore a determining factor for the
landuse pattern. Irrigation is only possible on areas equipped for irriga-
tion which is implemented based on Dll and Siebert (2000).
Production costs are derived from the GTAP database (Narayanan
and Walmsley, 2008) and include factor costs for labor, capital, and
 A. Biewald et al. / Ecological Economics 101 (2014) 4353
Fig. 1. Simplied MAgPIE ow chart of key processes. Economic regions in MAgPIE: AFR = Sub-Sahara Africa, CPA = Centrally Planned Asia (incl. China), EUR = Europe (incl. Turkey),
FSU = Former Soviet Union, LAM = Latin America, MEA = Middle East and North Africa, NAM = North America, PAO = Pacic OECD (Australia, Japan and New Zealand), PAS = Pacic
Asia, SAS = South Asia (incl. India).
intermediate inputs. Investments in technological change allow MAgPIE
to increase crop yields in a particular region. The productivity of the ir-
rigated and the rainfed production are both driven by the same regional
technical change factor. The endogenous implementation of technolog-
ical change (TC) is based on a surrogate measure for agricultural land
use intensity (see details in Dietrich et al. (2012)). Expansion of crop-
land is the alternative to increase the production level. The expansion
involves land-conversion costs for every unit of cropland, which ac-
count for the preparation of new land and basic infrastructure invest-
ments (Krause et al., 2013). Land-conversion costs are based on
country-level marginal access costs generated by the Global Timber
Model (GTM) (Sohngen et al., 2009).
Finally, the special feature of MAgPIE_trade is the inclusion of bilat-
eral trade between each of the world regions instead of using self-
sufciency rates as in the general MAgPIE version (Biewald et al.,
2011). For this, international trade and transport costs for each traded
unit are added to the objective function. The costs for bilateral transport
for each commodity are derived by calculating the difference between
bilateral export values at FOB prices (Free On Board) and bilateral im-
port values at CIF prices (Cost Insurance and Freight) and dividing it
by the traded quantity. Since the data are based on the GTAP database
(Narayanan and Walmsley, 2008) and GTAP reports all quantities in
US$ values, we had to divide the trade volume by prices taken from
FAO to get the traded quantity (FAO, 2010). In a similar manner, the
trade barriers are calculated. Import duties are derived as the difference
between bilateral import values at market prices and at world prices,
whereas export duties are derived as the difference between bilateral
export values at world prices and at market prices. As with transport
costs, we divide the results by the traded quantities to obtain trade
margins per ton.
2.2. Model Description of LPJmL
Vegetation growth, crop yields and water consumption are calcu-
lated with the global, dynamic and process-based model LPJmL. Natural
Table 1
Production activities in MAgPIE.
Categories Production activities
Cereals Temperate cereals; maize; tropical cereals; rice
Oilcrops Soybean; rapeseed; groundnut; sunower
Oilpalm Oilpalm
Roots/pulses Pulses; potato; cassava
Sugar Sugarbeet; sugarcane
Veg/fruits Vegetables, fruits, nuts (one category)
Other Fodder; ber
Livestock Ruminant meat; pork; chicken; eggs; milk
45
vegetation is represented at the biome level by nine plant functional
types (Sitch et al., 2003) and growth, phenology and harvest of 12
crop functional types (CFTs) account for the agricultural production
(Bondeau et al., 2007). The model ensures a global balance of carbon
uxes (gross primary production, auto- and heterotrophic-respiration),
pools (in leaves, sapwood, heartwood, storage organs, roots, litter and
soil), and water uxes (interception, evaporation, transpiration, soil
moisture, snowmelt, runoff, discharge) (Rost et al., 2008). Processes of
carbon assimilation and water consumption are parametrized on the
leaf level and scaled to the ecosystem level. Carbon and water dynamics
are closely linked so that the effects of changing temperatures, declining
water availability and rising CO2 concentrations are accounted for and
their net effect can be evaluated (Gerten et al., 2004, 2007). Physiologi-
cal and structural plant responses determine water requirements and
water consumption and thereby the virtual water content (VWC) of
harvested crops. Due to the source of the water used for plant produc-
tion, the consumption of blue and green water on irrigated and rainfed
cropland can be distinguished. All processes are simulated at a daily res-
olution and on a global grid of 0.5. Simulations were carried out with
the option of growing each crop in each grid cell so that the current
land use pattern is not applied. Thus, the redistribution of crop produc-
tion under changing climatic conditions into previously unsuitable
regions can be captured as well as the loss of formerly productive
areas. For irrigated potential production, the water requirement of the
crops is fullled by adding sufcient irrigation water to the available
precipitation of the respective grid cell. Thereby, virtual water contents
on rainfed and irrigated areas can be derived for the potential crop pro-
duction. The suitability of the model (and its predecessor LPJ without
cropland) for vegetation/crop and water studies has been demonstrated
by validating simulated phenology and yields (Bondeau et al., 2007),
river discharge (Biemans et al., 2009; Gerten et al., 2004), soil moisture
(Wagner et al., 2003), evapotranspiration (Gerten et al., 2004; Sitch
et al., 2003) and irrigation water requirements (Rost et al., 2008).
2.3. Model Application
We describe the methodology used to evaluate the impact of trade
on local blue water scarcity. Our approach identies grid-cells where
blue water is scarce and in which water is saved through international
imports or used for the production of crops for export.
2.3.1. Extracting the Net Trade Effect
Using MAgPIE_trade, average landuse patterns for the years 2000
until 2010 are derived for the different food crops (see Table 1 for a
closer specication) on a spatial grid with a 0.5 degree resolution.
Since all simulation results refer to values of the year 2005 we omit,
46 A. Biewald et al. / Ecological Economics 101 (2014) 4353
for reasons of comprehensibility, the year. Agricultural production
is distributed on rainfed areas (without irrigation) and on irrigated
areas with a limited supply of blue water for irrigation. In order to
extract the impact of trade on crop production, either by reducing pro-
duction due to imports, or due to additional production for exports, two
simulations are performed: First, a simulation with bilateral trade
constrained only by transport costs and trade barriers and second, a
simulation where trade is disabled enforcing self-sufciency of each re-
gion. For each crop and in every grid, the magnitude of the difference in
production between the simulations provides us with the desired infor-
mation on the impact of trade on the production in these grid-cells
(Eqs. (1), (2)).
crop crop crop
P I P nt I P wt I crop
crop crop crop
P R P nt R P wt R crop:
Pcrop [tDM] is the trade-related change for irrigated (Pcrop I )
and rainfed (Pcrop
R ) crop production which is derived by the difference
of the respective production when trade is disabled (index nt) and
when trade is enabled (wt). A negative value of Pcrop indicates that
production of the specic crop in this grid-cell is higher than would be
necessary to fulll the demand of this region and is therefore used for
export. A positive value indicates that production is spared through
trade.
When extracting the effect of trade in this way, one has to consider
that the global production in the two simulations is not necessarily
identical and that therefore the total difference in produced crops
does not have to be zero. The human regional demand for food (vegetal
as well as animal) is xed beforehand according to the population and
the GDP. But the value of crops for the production of livestock, necessary
to fulll the human food demand, may vary. By extracting the trade
effect, we look at the effect of shifting the livestock production from ex-
port regions to import regions. Since conversion efciencies from feed
crops to livestock products differ for each region (Fig. 2), the difference
in the amount of produced crops between the two simulations cannot
be zero. Feed conversion efciencies are calculated by a feed energy
balance model (Weindl et al., 2010) which connects production of the
ve livestock commodities (Table 1) with their specic feed energy re-
quirements (consistent with estimates by Wirsenius (2000)) to the re-
gional supply of different feed categories. Supply of primary crops and
conversion byproducts is derived from FAO Food Balance Sheets and
supplemented by estimates on production and use of non-marketable
feed.
Summing up the differences between the simulations over the grid-
cells in each region, we can extract the net trade effect on total food crop
production for each crop and region. In order to be able to aggregate and
compare the trade effect for the different crops, we converted them into
Joule. On the regional level we are able to estimate three factors which
are responsible for the net trade effects on production: the regional
Fig. 2. Livestock production efciencies in the 10 MAgPIE world regions.
food crop exports, the regional food crop imports and the food crop de-
mand for livestock. The net trade effect related to livestock can be eval-
uated as balance of export and import minus the difference in the
livestock demand.
2.3.2. Trade Effects on Agricultural Water Use
The model LPJmL provides for 2005, for every cell and every crop a
virtual blue and green water content. Multiplying the virtual water con-
tent of each crop with the trade-related production of each crop, we ob-
tain cell-specic information about the consumption of blue and green
water related to trade. Positive values denote water savings through
trade and negative values indicate water use for export.
The changes in virtual water usage are calculated for green and blue
1 water separately:
X
15
crop crop
X
15
crop crop
2 VWU G P I  VWC IG P R  VWC RG 3
crop1 crop1
X
15
VWU B
crop
P I
crop
 VWC IB : 4
crop1
VWU [m3 yr1] is the trade-related change in virtual water usage for
green (VWUG) and blue (VWUB) water, Pcrop [tDM] is the trade-
related change for irrigated (Pcrop
I ) and rainfed (Pcrop
R ) crop produc-
tion in tons dry matter (=DM), VWCcrop [m3 tDM1 yr1] is the virtual
water content for each crop, for rainfed green (VWCcrop
RG ), irrigated green
(VWCcrop crop
IG ) and irrigated blue (VWCIB ) virtual water consumption. The
rainfed blue virtual water content is per denition zero and does there-
fore not show up in VWUB.
2.3.3. Water Scarcity
In order to quantify the signicance of trade related water usage,
two factors have to be considered: the scarcity of the available water
itself and the change in consumption induced by trade. With the agro-
economic approach of MAgPIE_trade, scarcity for the year 2005 is eval-
uated by using the water shadow price WSPB [US$ m3]. The marginal
value of irrigation water generates the price for an additional unit
of water in the respective cell in the context of the global optimization
process and, therefore, indicates the scarcity of water per grid cell.
2.4. Trade-related Value of Blue Water
The evaluation of trade-related water use is achieved by multiplying
local water scarcity as indicated by the blue water shadow price WSPB
for the year 2005 with the trade related blue water use changes VWUB
for 2005 resulting in the water shadow price weighted trade-related
value of blue water use TVB [US$ yr1] for the year 2005 (Eq. (5)).
TV B VWU B  WSP B : 5
The trade-related value of blue water may have two signs: grid-cells
with negative values are water scarce and blue virtual water is exported,
and grid-cells with positive values are water scarce and blue water is
saved through trade. Grid-cells without irrigated production or with a
water shadow price of zero are not taken into account.
3. Results
Globally, trade of food crops, fodder and livestock reduces caloric
crop production by 15%; this reduction is the result of different regional
livestock conversion efciencies. The region where trade inuences
agricultural production most is Africa (reducing production by 121%),
whereas trade in Europe spares only 3%.
More world regions are net importers than net exporters. In Central-
ly Planned Asia, the Former Soviet Union, Africa, South Asia and Middle
 A. Biewald et al. / Ecological Economics 101 (2014) 4353
Fig. 3. Quantitative contribution of livestock/feed imports, food crop imports and food
crop exports to the net trade effect. The feed effect is the trade-related calory demand
for the production of livestock.
East trade results in a reduction of crop production by more than
1000 PJ (Figs. 3, 4), whereas in Europe, Latin America, North America
and Pacic OECD the net trade effect leads to changes in production
between 500 and 500 PJ, only Pacic Asia has net exports of more
than 1000 PJ.
In Fig. 3 the net trade effect is split into its driving factors: trade-
related crop production for import and export and trade-related pro-
duction for feed. Here one can see that Africa, Centrally Planned Asia,
Middle East and South Asia are net importers of food crops, whereas
the Former Soviet Union and Europe are net importers of fodder and
livestock. Only Latin America is a net exporter of fodder (specically
oilcrops and sugar) and a net importer of food crops (mainly cereals).
Globally, the crop category most inuenced by trade is cereals
(Fig. 5) (36% reduction of production), followed by roots & pulses
(17%). In contrast, trade leads to hardly any net change in the produc-
tion of rice and vegetables & fruits. The trade-related net changes in pro-
duction for the crop categories and the single regions can be found in
the supplement of this article.
3.1. The Trade-effect on Irrigated and Rainfed Production
Trade reduces global rainfed crop production by 20% and leads to an
extension of global irrigated production of 13%. In Africa, South Asia,
Middle East and Latin America, trade reduces production on rainfed
areas and extends it on irrigated areas (Fig. 4).
Oilpalm is the crop with the highest trade-related reduction of irri-
gated production (125%). Rice, oilcrops, sugar and roots & pulses are
crops where trade leads to more irrigated production (Fig. 5).
In the model MAgPIE_trade, changes in production can be achieved
by two measures, through investments in productivity and through ex-
pansion of agricultural land. Trade has therefore also an impact on both,
Fig. 4. Net trade effect on rainfed and irrigated production for the different regions: nega-
tive numbers denote production for exports; positive numbers denote production not
necessary due to imports.
47
Fig. 5. Net trade effect on rainfed and irrigated production for the crop categories: negative
numbers denote production for exports; positive numbers denote production not neces-
sary due to imports.
it reduces global production area by 26% (Fig. 6), but has a mixed effect
on global irrigated and rainfed yield (Fig. 7). The difference in yield
productivity between irrigated and rainfed production results from
agricultural land expansion, trade reduces global rainfed area by 40%,
but extends irrigated area by 27%.
The net trade effect in terms of agricultural rainfed land is highest in
Africa (187%) with a small impact on yields. Trade leads to almost no ex-
pansion of agricultural land in Centrally Planned Asia and Europe and to
a reduction of crop land in Pacic OECD and North America. Middle East
is the region where trade reduces the pressure on irrigated production
most, while at the same time saving 27% of agricultural, rainfed area.
In Latin America and Pacic Asia, trade leads to an additional use of irri-
gated land by 58% and 68%, respectively. In the Former Soviet Union
trade reduces agricultural rainfed land (70%) and relieves the pressure
on productivity.
3.2. Trade Effect on Water Use
Trade leads to global water savings of 949 Gm3 yr1, which is 17% of
the water consumed globally for crop production. This water saving
consists largely of green water. Through trade 18% (932 Gm3 yr 1)
green water and 5% (17 Gm3 yr1) blue water is saved in crop produc-
tion (Fig. 8).
Africa, Middle East and South Asia save through trade green and
total water of more than 200 billion Gm3 yr1. The region with the
highest relative (100%) and absolute (386 Gm3 yr1) net water savings
is Africa. Europe, Latin America, North America, Pacic OECD and Pacic
Asia increase their total and green water use through trade. The highest
blue water savings through trade take place in South Asia (39%) and
here especially in India and Pakistan (Fig. 9). In contrast almost no
blue water is saved in North America, the Former Soviet Union and
Pacic OECD. Pacic Asia and Southern Europe (Portugal, Spain, Italy,
Fig. 6. Net trade effect on rainfed and irrigated land expansion for the different regions:
negative numbers denote land expansion for export production; positive numbers denote
agricultural land not necessary due to imports.
48 A. Biewald et al. / Ecological Economics 101 (2014) 4353
Fig. 7. Net trade effect on average rainfed and irrigated productivity: negative numbers
denote increased pressure on productivity due to trade; positive numbers denote pro-
ductivity increases not necessary due to imports.
Greece and Turkey) are the regions where trade leads to negative blue
water savings of 64% and 32%, respectively.
Trade leads to the highest savings in total and green water con-
sumption for the production of cereals (Fig. 10). A small negative
net saving for green water is obtained in the case of sugar production.
Water savings for irrigated production do not equal blue water savings
(see e.g. rice, oilpalm and roots and pulses). A detailed summary of
regional trade-related water savings for each crop-category can be
found in the supplement.
The global net savings of blue water are rather small (17 Gm3 yr1,
5%). But the positive regional gross savings of virtual blue water
of 93 Gm3 yr 1 show that some regions save large amounts of blue
water. The specic negative and positive blue water savings per grid
cell can partly be very high (Fig. 9).
3.3. Water Scarcity
The gridded water shadow price (Fig. 11) only has values in cells
which have area equipped for irrigation and in which blue water is
scarce. The highest local water shadow prices can be found in Central
Asia. In South Asia a large area is irrigated and water is scarce, whereas
the large irrigated area in Latin America is only partly water scarce. The
western part of North America is irrigated and blue water is scarce, in
Europe agricultural area is almost everywhere equipped for irrigation
but only in Southern Europe water is scarce.
3.4. The Trade-related Value of Blue Water
In order to receive the trade-related value of blue water TVB (Fig. 12)
we multiply on the grid-cell level the water shadow prices WSPB
(Fig. 11) with the trade-related changes in blue water (Fig. 9). The global
Fig. 8. The net trade effect on regional, virtual water consumption: negative numbers de-
note water consumption for export production; positive numbers denote water savings
due to imports.
trade-related net value of blue water amounts to 2451 106US$ and is
the sum of the positive gross value of 7014 106US$ and the negative
gross value of 4563 106US$ (Table 2). The regions which contribute
most to the positive gross value are Middle East and South Asia, whereas
only Europe considerably contributes to the negative net values. The re-
gions Latin America and Pacic OECD are the regions with the smallest
net values of TVB of 9 106US$ and 2 106US$, respectively.
The highest positive values of TVB (water is scarce and these coun-
tries import virtual water) can be found in India, Morocco, West
Africa, northern China, parts of North America and Mexico. The highest
negative values of the index (water is scarce and the countries are
exporters of virtual water) can be found in Portugal, Spain, Turkey and
Afghanistan.
The net values of TVB for the different crop categories (see Table 3)
are relatively small compared to the negative and positive gross values.
For example, cereals exhibit a net value of 413 106US$, while the
negative gross value is 2203 106US$ and the positive gross value
2616 106US$. The highest positive net value of TVB is attributed to
vegetables & fruits, the lowest value to rice.
Local extreme values for TVB can be found for cereals and sugar in
India and Morocco (positive) and for sugar and vegetables & fruits in
Spain, Portugal and India (negative). For the global values of each crop
category see Table 3, for cell based maps on the TVB for the crop catego-
ries look at the supplement.
While the global area on which blue water is saved through or used
for the trade of agricultural goods is relatively large, the area where
water is scarce and trade-related changes of blue water take place are
much smaller. Large parts of the Former Soviet Union, Centrally Planned
Asia, northern Europe and Africa save water through trade, but they do
not show up in the TVB. While Latin America, West Africa and Pacic
OECD use water for the export which is also not relevant in the TVB.
4. Discussion
The contribution of the concept of virtual water trade to the compre-
hension of local water scarcity is currently under debate (Biewald,
2011; Biewald and Rolinski, 2012). Employing three steps, our study ad-
vances the applicability of the virtual water concept. First, it estimates
trade-related changes in food crop production. Through the compre-
hensive modeling approach used in this study, we are able to determine
not only the direct trade-effect on the production of food crops, but also
the indirect trade-effect on crops originating from trading livestock.
Second, employing the results on trade-related production changes,
the study determines green and blue water savings due to crop trade
on a grid-cell level. Third, the study combines on a subnational level
trade related blue water changes with an agroeconomic water-scarcity
indicator in order to identify water scarce grid-cells which produce
export crops or do not need to produce crops due to imports.
4.1. Trade-related Changes in Production
Although the regional food demand is xed in the model, due to
the regionally different livestock production efciencies, trade leads
to a reduction in global crop production and area. The trade-related
land savings are approximately twice as high (825 105 ha) than in a
study by Fader et al. (2011) (e.g. 410 105 ha), which also considers
the effect of crop trade, but without taking feed into account. This effect
is considerable due to a share of 33% of feed in global cereal production
(Steinfeld et al., 2006).
While trade reduces total production and area, it increases irrigated
production and area. There are two reasons for the increase in irrigated
areas in the trade scenario. First, through the unequal distribution
of irrigated areas over the different regions, local production which is
compensating irrigated exports in the self sufciency case might have
to be produced on rainfed sites due to the lack of irrigated and more pro-
ductive area. Second, trade permits export production under optimal
 A. Biewald et al. / Ecological Economics 101 (2014) 4353
Fig. 9. The net trade effect on blue water consumption in 2005 on a 0.5 grid basis: blue cells indicate water savings through trade; red cells indicate water used for exports.
conditions, demanding crops in terms of water (which are often also
high added value products) can therefore be produced on the most ad-
vantageous, namely irrigated sites. Since with no trade every region has
to produce everything for themselves, it might be optimal to leave irri-
gated areas unused and produce less water demanding crops on more
productive, but not irrigated areas.
Trade affects the net importers either through freeing up agricultural
land (Africa, South Asia) or through the ability to reduce pressure on
productivity (Centrally Planned Asia, Middle East) or because of both
(Former Soviet Union). Europe and North America are net exporters
of food crops, but import an almost equal amount of crop calories in
form of fodder or livestock, so that net changes in productivity and
land are rather small. Trade increases global irrigated production, main-
ly because of the two exporting regions, Pacic Asia and Latin America.
For the major exporter Pacic Asia, the trade-related expansion of irri-
gated areas has two reasons: First, the irrigated production for the
export of rice and, to a lesser extent, cereals, oilcrops, roots & pulses
and vegetables & fruits and, second, the shift of rice production from
very productive and climatic advantageous areas (but which are not
equipped for irrigation) to cells which are less productive (but can be
irrigated) in order to free up area for the export production of the
water productive oilpalm. The decisive factor for production in this
case is therefore productivity and not water availability.
In Latin America, cereal imports (used as fodder) are compen-
sated on rainfed areas, but production of sugar and oilcrops for the ex-
port takes place on irrigated area. This trade-related de-allocation of
Fig. 10. Blue, green, irrigated and rainfed trade-related water savings for each crop
category.
49
irrigated area in Latin America is due to the fact that the production of
sugar and oilcrops is water intensive, and that it is therefore of compar-
ative advantage to grow them on irrigated areas and to grow the less
water intensive cereals on only rainfed areas. We can therefore not con-
rm that exporters produce mostly on rainfed areas, while importers
would have to produce irrigated (Aldaya et al., 2010).
4.2. Blue and Green Water Savings
The total global water savings in our study amount to
949 Gm3 yr1, while Oki and Kanae (2004) calculate total water savings
of 455 Gm3 yr1 and Chapagain et al. (2006) of 352 Gm3 yr1. The fact
that our total water savings are approximately twice as high than in
these studies shows that multiplying an average virtual water content
with the data of imported crops cannot cover all the aspects of trade
such as different regional livestock efciencies. The domination of vir-
tual green water trade over blue water trade is well established in the
literature (Fader et al., 2011; Hanasaki et al., 2010; Yang et al., 2006)
and can also be shown with our results.
The amount of exported or imported crops in the different regions is
not necessarily proportional to the amount of virtual water traded. The
greatest net exporter, Pacic Asia, exports 1119 PJ of food crops, but in-
creases green water consumption only by 57 Gm3 yr1, while the live-
stock exporter Latin America exports only 222 PJ and increases net
water consumption by 141 Gm3 yr1. The reason for this discrepancy
is due to the difference in export crops and their virtual water contents.
While Latin America mainly exports sugar with a high virtual water con-
tent, Pacic Asia is a major exporter of oilpalm which has a low virtual
water content. On the other hand, the net importer Europe increases
total water consumption with trade because it exports cereals which
need a relative high amount of water per Joule, while importing crops
which need relatively less water like oilpalm and roots & pulses.
Net blue water savings are realized in Centrally Planned Asia, the
Former Soviet Union, Middle East and South Asia (Fig. 8). While for
the former two, trade-related net savings of irrigated production lead
to savings of blue water, in Middle East and South Asia blue water is
saved although trade increases produced calories on irrigated areas.
This antagonism can be explained by the more efcient use of blue
water in irrigation agriculture for export crops, while irrigated crops
compensating imports are produced inefciently and use therefore a
higher share of blue water.
The bilateral approach of estimating water savings, used in the liter-
ature, leads to difculties when trying to distinguish blue and green
50 A. Biewald et al. / Ecological Economics 101 (2014) 4353

Fig. 11. Water scarcity indicated by water shadow price in 2005 on a 0.5 grid basis.

water, since import and export countries cannot be assumed to use
water of the same origin. Using this approach de Fraiture et al. (2004)
estimate blue water savings of cereals. However, due to the lack of
specic information on the potential water usage of the importing
countries, the authors assumed that these countries use approximately
two and a half times more green than blue water for the production
of the imported cereals. The study obtains blue water savings of
112 Gm3 yr1 and green water savings of only 51 Gm3 yr1. That blue
water savings are higher than green water savings results from the dif-
ference of water productivities between importing and exporting coun-
tries and the assumption that importing countries compensate irrigated
imports with irrigated production. An additional imprecision results
from the national water productivities, these neglect local differences
in productivities, which is especially distorting for large countries with
different climatic zones.
Our study nds for cereals smaller blue water savings (31 Gm3 yr1)
and larger green water savings (744 Gm3 yr1) than de Fraiture et al.
(2004). With our grid-cell based approach we can determine water pro-
ductivity on a small subnational scale and we take into account that
there are local differences in crop production (irrigated or rainfed).
Our estimates for blue water savings are therefore always smaller
than the green water savings. That some crops (rice, oilcrops, soybean)
have globally trade-related negative blue water savings as well as posi-
tive green water savings results from that phenomenon that some
exporting countries produce with irrigation, while the importing coun-
tries can only produce on rainfed areas.
We compare our global water saving estimates for different crops
with other studies in the supplement. Since our study includes not
only food crop related water savings, but also indirect savings due to
livestock production, our global water savings should be systematically
higher for feed crops. This is conrmed by our results; our water savings
of the major feed crops temperate cereals and maize are clearly higher
than the ones found in the literature. For oilcrops, roots & pulses and
soybean we are well in the range of the literature.
4.3. Water Shadow Price
Prices for irrigation water are often a political decision and do not in-
dicate the economic value of blue water. There are two reasons for this
misvaluation. First, the real scarcity of blue water is not known, since

Fig. 12. Trade-related value of blue water TVB in 2005 on a 0.5 grid basis, product of water shadow price (Mio US$ m3 yr1) and blue water consumption due to trade (m3 yr1): green cells
indicate water savings on water scarce site through trade; orange cells indicate water used for the export on water scarce sites.
 A. Biewald et al. / Ecological Economics 101 (2014) 4353 51

Table 2 is negatively correlated to the low blue water savings (9 Gm3 yr1).
Regional trade-related value of blue water in 106US$. This means that although only little water is saved the region prots
Regions Net values Neg. gross values Pos. gross values enormously from trade. This discrepancy in water savings and TVB re-
sults amongst others from sugar and cereal imports to Morocco and
AFR 45 52 7
CPA 160 9 169 Egypt, which save relatively little water, but due to the lack of any
EUR 3157 3231 74 other possibilities has to be compensated on very water scarce sites.
FSU 48 1 49 Due to the small irrigated area in Middle East compared to total area,
LAM 9 37 29
the visible blue water savings are small. If irrigated area could be ex-
MEA 2246 118 2364
NAM 151 154 3 tended, the TVB would probably be even more extreme.
PAO 2 9 11 All other regions have TVB values which are much smaller than the
PAS 60 60 0 ones from Europe, South Asia and Middle East. While Centrally Planned
SAS 3416 893 4309 Asia is the region with the second highest blue water savings after South
GLO 2451 4563 7014
Asia, the TVB with 160 106US$ is relatively small, resulting from the
low local water scarcity and the fact that the water savings through
the import of cereals are balanced out through the export of roots &
e.g. there are no information on the size of the ground water reser- pulses. In Latin America and Pacic Asia trade impacts blue water
voirs, and second, it might be a political decision to support local usage considerably, but since the water shadow prices are very low
farmers by subsidizing irrigation water (Chapagain and Hoekstra, or even zero, the trade-related value of blue water changes is very
2008). On the other hand, the consumption of blue water might small. Since in Africa only few sites are equipped for irrigation, the
have external effects on the environment which are not reected in trade related changes in blue water consumption and the TVB are
our water shadow price, but might be part of the politically deter- small. Spendings on virtual blue water results from the export of roots
mined water price. The wide range of irrigation water prices in & pulses from South Africa and Uganda. In North America a small posi-
Europe reects this complexity of price determining factors. Prices tive value of TVB corresponds to small blue water savings, which results
for irrigation water in Europe range from heavily subsidized irrigation from the fact that on the water scarce areas export production of cereals
water of 0.007 m3 (0.009 US$ m3) in Bulgaria to 1.04 m 3 compensates oilcrop imports.
(1.36 US$ m3) for water charged with environmental taxes in the The crops which are responsible for the highest global blue water
Netherlands (a comprehensive literature review on irrigation prices savings (Fig. 10) are not the ones responsible for the highest global
can be found in Berbel et al. (2007)). Since politically determined irri- values of TVB (Table 3). While cereals and oilpalm are the crop catego-
gation water prices cannot be compared to our results, we have to use ries which save most blue water through trade, vegetables & fruits is
similar price estimations for comparison, which until now are rare. the group with the highest positive TVB of 1961 106US$. While for ce-
Novo et al. (2009) summarize different studies for Spain, which result reals the exports from water scarce sites are partly neutralized by com-
in water shadow prices of 0.150.3 m 3 (0.190.39 US$ m3) pensating imports, oilpalm is mostly grown on not water scarce sites.
for 2005, which t well in our national estimate of 0.10.3 m 3 Vegetables & fruits on the other hand, are grown on very water scarce
(0.130.39 US$ m3). sites in India and exhibit therefore a high value of TVB. The high positive
and negative global gross values for crops also show that irrigated crops
4.4. Valuing the Trade Related Blue Water Changes are exported from and imported to water scarce regions at the same
time.
The value of the trade-related blue water changes (TVB) adds infor- The variability of the range of the TVB values on grid-cell level
mation on site-specic water scarcity to the cellular blue water savings. depends on the different crops, the water scarcity of the cell and the
The regions with the highest absolute values of TVB are Europe, Middle amount of crops imported or exported (see supplement). While the
East and South Asia (Table 2), but in each of these regions the high extreme grid-cell values of TVB of all crops have a maximum value
value of TVB has to be interpreted differently. of 45 106US$ and a minimum value of 30 106US$, the highest
Europe is a large exporter of blue water (19 Gm3 yr1) and has a cell value of TVB is 158 106US$ for sugar and the lowest cell value
high negative value of TVB (3157 106US$), which shows that water is 73 106US$ for cereals. This shows that exports and imports of
intensive export goods are produced on water scarce sites. In particular, different crops in the same water scarce grid-cell can neutralize each
the high TVB results from exporting irrigated cereals and sugar from other. The country where this neutralization of positive and negative
water scarce sites in Spain, Portugal, Italy and Turkey. South Asia has values of TVB is most apparent is India, where the imports of cereals
the highest value of TVB (3416 106US$) and the highest amount of and vegetables & fruits lead to positive values of TVB, whereas the
blue water savings (39 Gm3 yr1), alleviating its local water scarcity exports of roots & pulses and oilcrops lead to negative TVB values.
by importing water intensive goods. Here, India is the country mainly re-
sponsible for the high TVB since it imports water intensive cereals, vege- 5. Conclusion
tables & fruits, which it otherwise would have to produce on water
scarce sites. Middle East has a high value of TVB (2246 106US$) which With our study we could show that local water scarcity is alleviated
through imports of agricultural goods in parts of India, Morocco, Egypt
and Pakistan, as well as intensied through exports in parts of Turkey,
Table 3
Spain, Portugal, Afghanistan and the US.
Trade-related value of blue water in 106US$ for the different crop categories. For the countries where trade aggravates water scarcity, exporting is
of comparative advantage because the price of irrigation water does not
Category Net values Neg. gross values Pos. gross values
depict its scarcity. This is true in the model setup as well as in reality. For
Cereals 413 2203 2616 example, groundwater is often overexploited when water is scarce on
Rice 379 620.6 242
productions sites. In such cases as well as in the case of overcharging en-
Oilcrops 41 576 617
Oilpalm 918 11 930 vironmental ows, costs for the environment and for future generations
Roots & pulses 291 365 74 are not internalized.
Sugar 213 2333 2120 The ndings reecting the positive as well as the negative impacts of
Vegetables & fruits 1961 1490 3451 trade on scarce water resources are extremely important for policy
All 2451 7598 10,048
makers, since they show how international trade inuences local
52 A. Biewald et al. / Ecological Economics 101 (2014) 4353
water situations. When implementing policy measures which impact
global trade such as trade liberalization, trade barriers or agricultural
subsidies, policy makers have to take into account the indirect effect
of such measures as well as the direct effect.
The importance of the ndings of this paper and the concept of vir-
tual water trade holds true despite critical discussions about the concept
initiated by Dennis Wichelns (2010a,2010b, 2011). The major point of
the criticism comprises the fact that the concept is not able to tell if
trade of virtual water is caused by local water conditions or due to the
scarcity or afuence of other important resources such as capital, labor
or land. While this is true this criticism does not diminish the meaning-
fulness of this study, since it provides important insights into the rela-
tions between trade and local water scarcity.
Furthermore, as discussed in Biewald (2011) and Biewald and
Rolinski (2012), estimating trade-related virtual water savings can be
useful, but only if relevant for water scarcity and analyzed on a sub-
national level. National water footprints or water savings without de-
termining the origin are useless, since they cannot disclose the effect
water scarcity.
Acknowledgments
We are grateful to Holger Hoff for insightful comments, to Jan
Kowalewski for his support in creating the graphics and to Markus
Bonsch for fruitful discussions on the subject. We received funding
from the BMBF in the context the project Sustainable water manage-
ment in a globalized world (01UN1009) and from the European
Union in the context of the project CARBO-Extreme (FP7/226701).
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.ecolecon.2014.02.003.
References
Alcamo, J., Dll, P., Henrichs, T., Kaspar, F., Lehner, B., Rosch, T., Siebert, S., 2003. Global es-
timates of water withdrawals and availability under current and future business-
as-usual conditions. Hydrol. Sci. J. 48, 339348. http://dx.doi.org/10.1623/hysj.
48.3.339.45278.
Aldaya, M.M., Allan, J.A., Hoekstra, A.Y., 2010. Strategic importance of green water
in international crop trade. Ecol. Econ. 69, 887894. http://dx.doi.org/10.1016/
j.ecolecon.2009.11.001.
Allan, J.A., 1996. Water use and development in arid regions: environment, economic de-
velopment and water resource politics and policy. Rev. Eur. Community Int. Environ.
Law 5, 107115. http://dx.doi.org/10.1111/j.1467-9388.1996.tb00258.x.
Berbel, J., Calatrava, J., Garrido, A., 2007. Water pricing and irrigation: a review of the
European experience. Irrigation Water Pricing Policy the Gap Between Theory and
Practice 295327.
Biemans, H., Hutjes, R., Kabat, P., Strengers, B., Gerten, D., Rost, S., 2009. Impacts of precip-
itation uncertainty on discharge calculations for main river basins. J. Hydrometeorol.
10, 10111025.
Biewald, A., 2011. Give virtual water a chance! An attempt to rehabilitate the concept.
GAIA Ecol. Perspect. Sci. Soc. 20, 168170.
Biewald, A., Rolinski, S., 2012. The theory of virtual water: why it can help to understand
local water scarcity. GAIA Ecol. Perspect. Sci. Soc. 12, 8890.
Biewald, A., Rolinski, S., Lotze-Campen, H., Schmitz, C., 2011. The effect of oil price in-
creases on agricultural trade: simulations with a global landuse model. Conference
Paper, Presented at the EAAE in Zuerich.
Bondeau, A., Smith, P.C., Zaehle, S., Schaphoff, S., Lucht, W., Cramer, W., Gerten, D., 2007.
Modelling the role of agriculture for the 20th century global terrestrial carbon
balance. Glob. Chang. Biol. 13, 679706.
Brook, A., Kendrick, D., Meeraus, A., 1988. GAMS, a user's guide. ACM SIGNUM Newsl. 23.
Chapagain, A.K., Hoekstra, A.Y., 2008. The global component of freshwater demand and
supply: an assessment of virtual water ows between nations as a result of trade
in agricultural and industrial products. Water Int. 33, 1932. http://dx.doi.org/
10.1080/02508060801927812.
Chapagain, A.K., Hoekstra, A.Y., Savenije, H.H.G., 2006. Water saving through international
trade of agricultural products. Hydrol. Earth Syst. Sci. 10, 455468.
de Fraiture, C., Cai, X., Amarasinghe, U., Rosegrant, M., Molden, D., 2004. Does international
cereal trade save water?: the impact of virtual water trade on global water use. IWMI
Research Reports H035342. International Water Management Institute.
Dietrich, J.P., Schmitz, C., Mller, C., Fader, M., Lotze-Campen, H., Popp, A., 2012. Measuring
agricultural land-use intensity a global analysis using a model-assisted approach.
Ecol. Model. 232, 109118. http://dx.doi.org/10.1016/j.ecolmodel.2012.03.002.
Dll, P., Siebert, S., 2000. A digital global map of irrigated areas. Icid J. 49, 5566.
Fader, M., Gerten, D., Thammer, M., Heinke, J., Lotze-Campen, H., Lucht, W., Cramer, W.,
2011. Internal and external greenblue agricultural water footprints of nations,
and related water and land savings through trade. Hydrol. Earth Syst. Sci. 15,
16411660. http://dx.doi.org/10.5194/hess-15-1641-2011.
Falkenmark, M., Lannerstad, M., 2010. Food security in water-short countries coping
with carrying capacity overshoot. In: Martnez-Cortina, L., Garrido, A., Lpez-Gunn,
E. (Eds.), Re-thinking Water and Food Security. CRC Press, pp. 322.
Falkenmark, M., Rockstrm, J., 2004. Balancing Water for Humans and Nature: The New
Approach in Ecohydrology. Earthscan.
Falkenmark, M., Lundqvist, J., Widstrand, C., 1989. Macro-scale water scarcity requires
micro-scale approaches. Aspects of vulnerability in semi-arid development. Nat.
Res. Forum 13, 258267 (PMID: 12317608).
FAO, 2010. Faostat food and agriculture organization of the United Nations Statistics
Division. Last Access: 11/07/2010 http://faostat.fao.org/.
Garrido, A., Llamas, M.R., Varela-Ortega, C., Novo, P., Rodriguez-Casado, R., Aldaya, M.M.,
2010. Water Footprint and Virtual Water Trade in Spain: Policy Implications.
Springer.
Gerten, D., Schaphoff, S., Haberlandt, U., Lucht, W., Sitch, S., 2004. Terrestrial vegetation
and water balance hydrological evaluation of a dynamic global vegetation model.
J. Hydrol. 286, 249270.
Gerten, D., Schaphoff, S., Lucht, W., 2007. Potential future changes in water limitations of
the terrestrial biosphere. Clim. Chang. 80, 277299.
Government Ofce for Science, London, 2011. Foresight. The Future of Food and Farming.
Hanasaki, N., Kanae, S., Oki, T., Masuda, K., Motoya, K., Shirakawa, N., Shen, Y., Tanaka, K.,
2008. An integrated model for the assessment of global water resources part 2: appli-
cations and assessments. Hydrol. Earth Syst. Sci. 12, 10271037.
Hanasaki, N., Inuzuka, T., Kanae, S., Oki, T., 2010. An estimation of global virtual water ow
and sources of water withdrawal for major crops and livestock products using a
global hydrological model. J. Hydrol. 384, 232244. http://dx.doi.org/10.1016/
j.jhydrol.2009.09.028.
Hoekstra, A., Hung, P., 2005. Globalisation of water resources: international virtual water
ows in relation to crop trade. Glob. Environ. Chang. Hum. Policy Dimens. 15, 4556.
http://dx.doi.org/10.1016/j.gloenvcha.2004.06.004.
Hoekstra, A.Y., Mekonnen, M.M., 2012. The water footprint of humanity. Proc. Natl. Acad.
Sci. U. S. A. 109, 32323237. http://dx.doi.org/10.1073/pnas.1109936109.
Hoekstra, A.Y., Mekonnen, M.M., Chapagain, A.K., Mathews, R.E., Richter, B.D., 2012.
Global monthly water scarcity: blue water footprints versus blue water availability.
PLoS ONE 7. http://dx.doi.org/10.1371/journal.pone.0032688.
Islam, M.S., Oki, T., Kanae, S., Hanasaki, N., Agata, Y., Yoshimura, K., 2007. A grid-based
assessment of global water scarcity including virtual water trading. Water Resour.
Manag. 21, 1933. http://dx.doi.org/10.1007/s11269-006-9038-y.
Krause, M., Lotze-Campen, H., Popp, A., Dietrich, J., Bonsch, M., 2013. Conservation of un-
disturbed natural forests and economic impacts on agriculture. Land Use Policy 30,
344354.
Lotze-Campen, H., Mller, C., Bondeau, A., Rost, S., Popp, A., Lucht, W., 2008. Global food
demand, productivity growth and the scarcity of land and water resources: a spatially
explicit mathematical programming approach. Agric. Econ. 39, 325338.
Ma, J., Hoekstra, A.Y., Wang, H., Chapagain, A.K., Wang, D., 2006. Virtual versus real water
transfers within China. Philos. Trans. R. Soc. B 361, 835842. http://dx.doi.org/
10.1098/rstb.2005.1644 (PMID: 16767828 PMCID: 1609405).
Mekonnen, M.M., Hoekstra, A.Y., 2010. A global and high-resolution assessment of the
green, blue and grey water footprint of wheat. Hydrol. Earth Syst. Sci. Discuss. 7,
24992542.
Mekonnen, M.M., Hoekstra, A.Y., 2011. The green, blue and grey water footprint of crops
and derived crop products. Hydrol. Earth Syst. Sci. 15, 15771600.
Molden, D., 2007. Water for food, water for life: a comprehensive assessment of water
management in agriculture: summary. IWMI Books, Reports H039769International
Water Management Institute.
Narayanan, B., Walmsley, T., 2008. Global Trade, Assistance, and Production: The GTAP 7
Data Base. Center for Global Trade Analysis, Purdue University ed.
Novo, P., Garrido, A., Varela-Ortega, C., 2009. Are virtual water ows in Spanish grain trade
consistent with relative water scarcity? Ecol. Econ. 68, 14541464. http://dx.doi.org/
10.1016/j.ecolecon.2008.10.013.
Oki, T., Kanae, S., 2004. Virtual water trade and world water resources. Water Sci. Technol.
49, 203209.
Oki, T., Kanae, S., 2006. Global hydrological cycles and world water resources. Science 313,
10681072. http://dx.doi.org/10.1126/science.1128845.
Pingali, P., 2007. Westernization of Asian diets and the transformation of food systems:
implications for research and policy. Food Policy 32, 281298. http://dx.doi.org/
10.1016/j.foodpol.2006.08.001.
R. Development Core Team, 2011. R: A Language and Environment for Statistical Computing.
R Foundation for Statistical Computing, Vienna, Austria.
Ridoutt, B., Pster, S., 2010. A revised approach to water footprinting to make transparent
the impacts of consumption and production on global freshwater scarcity. Glob.
Environ. Chang. Hum. Policy Dimens. 20, 113120. http://dx.doi.org/10.1016/
j.gloenvcha.2009.08.003.
Rosegrant, M., Sombilla, M., 1997. Critical issues suggested by trends in food, popula-
tion, and the environment to the year 2020. Am. J. Agric. Econ. 79, 14671470.
http://dx.doi.org/10.2307/1244363.
Rost, S., Gerten, D., Bondeau, A., Lucht, W., Rohwer, J., Schaphoff, S., 2008. Agricultural
green and blue water consumption and its inuence on the global water system.
Water Resour. Res. 44, W09405. http://dx.doi.org/10.1029/2007WR006331.
Schmitz, C., Lotze-Campen, H., Gerten, D., Dietrich, J.P., Biewald, A., Bodirsky, B., Popp, A.,
2013. Blue water scarcity and the economic impacts of future agricultural trade and
demand. Water Resour. Res. 49, 36013617.
 A. Biewald et al. / Ecological Economics 101 (2014) 4353
Sitch, S., Smith, B., Prentice, I.C., Arneth, A., Bondeau, A., Cramer, W., Kaplan, J.O., Levis, S.,
Lucht, W., Sykes, M.T., Thonicke, K., Venevsky, S., 2003. Evaluation of ecosystem
dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic global
vegetation model. Glob. Chang. Biol. 9, 161185.
Sohngen, B., Tennity, C., Hnytka, M., 2009. Global Forestry Data for the Economic Modeling
of Land Use. Economic Analysis of and Use in Global Climate Change Policy. Routledge,
New York.
Steinfeld, H., Gerber, P., Wassenaar, T.D., Castel, V., Haan, C.D., 2006. Livestock's Long
Shadow: Environmental Issues and Options. Food & Agriculture Organisation.
van Vuuren, D., Isaac, M., Kundzewicz, Z., Arnell, N., Barker, T., Criqui, P., Bauer, N.,
Berkhout, F., Hilderink, H., Hinkel, J., Hochrainer, S., Hof, A., Kitous, A., Kram, T.,
Mechler, R., Scrieciu, S., 2009. Scenarios as the basis for assessment of mitigation
and adaptation. In: Hulme, M., Neufeldt, H. (Eds.), Making Climate Change Work
for Us ADAM Synthesis Book. Cambridge University Press.
Verma, S., Kampman, D.A., Sarg, P.v.D., Hoekstra, A.Y., 2008. Going against the ow:
a critical analysis of virtual water trade in the context of India's National River Linking
Program. Technical Report. UNESCO-IHE Institute for Water Education.
Vrsmarty, C., Green, P., Salisbury, J., Lammers, R., 2000. Global water resources: vul-
nerability from climate change and population growth. Science 289, 284288.
http://dx.doi.org/10.1126/science.289.5477.284.
53
Wagner, W., Scipal, K., Pathe, C., Gerten, D., Lucht, W., Rudolf, B., 2003. Evaluation of the
agreement between the rst global remotely sensed soil moisture data with model
and precipitation data. J. Geophys. Res. 108, 4611.
Weindl, I., Lotze-Campen, H., Popp, A., Bodirsky, B., Rolinski, S., 2010. Impacts of livestock
feeding technologies on greenhouse gas emissions. Contributed Paper at the IATRC
Public Trade Policy Research and Analysis Symposium.
Wichelns, D., 2010a. Virtual water: a helpful perspective, but not a sufcient policy criterion.
Water Resour. Manag. 24, 22032219. http://dx.doi.org/10.1007/s11269-009-9547-6.
Wichelns, D., 2010b. Virtual water and water footprints: policy relevant or simply
descriptive? Int. J. Water Resour. Dev. 26, 689695. http://dx.doi.org/10.1080/
07900627.2010.519533.
Wichelns, D., 2011. Virtual water and water footprints. Compelling notions, but notably
awed. GAIA Ecol. Perspect. Sci. Soc. 20, 171175.
Wirsenius, S., 2000. Human Use of Land and Organic Materials Modeling the Turnover
of Biomass in the Global Food System. Chalmers University, Gteborg, Sweden.
Yang, H., Reichert, P., Abbaspour, K.C., Zehnder, A.J.B., 2003. A water resources thresh-
old and its implications for food security. Environ. Sci. Technol. 37, 30483054.
http://dx.doi.org/10.1021/es0263689.
Yang, H., Wang, L., Abbaspour, K., Zehnder, A., 2006. Virtual water trade: an assessment of
water use efciency in the international food trade. Hydrol. Earth Syst. Sci. 10, 443454.