Article

pubs.acs.org/est

Water Flows in the Spanish Economy: Agri-Food Sectors, Trade and

Households Diets in an Input-Output Framework
Ignacio Cazcarro,* Rosa Duarte, and Julio Sánchez-Chóliz
Department of Economic Analysis, Faculty of Economics and Business, University of Zaragoza, Gran Vía, 2, 50005, Zaragoza, Spain
*
S Supporting Information

ABSTRACT: Seeking to advance our knowledge of water

flows and footprints and the factors underlying them, we apply,
on the basis of an extended 2004 Social Accounting Matrix for
Spain, an open Leontief model in which households and
foreign trade are the exogenous accounts. The model shows
the water embodied in products bought by consumers (which
we identify with the Water Footprint) and in trade (identified
with virtual water trade). Activities with relevant water inflows
and outflows such as the agrarian sector, textiles, and the agri-
food industry are examined in detail using breakdowns of the
relevant accounts. The data reflect only physical consumption,
differentiating between green and blue water. The results
reveal that Spain is a net importer of water. Flows are then
related to key trading partners to show the large quantities
involved. The focus on embodied (or virtual) water by activity
is helpful to distinguish indirect from direct consumption as
embodied water can be more than 300 times direct
consumption in some food industry activities. Finally, a sensitivity analysis applied to changes in diets shows the possibility
of reducing water uses by modifying households’ behavior to encourage healthier eating.

1. INTRODUCTION relationship between production, trade and water; so from

The Spanish economy has undergone profound structural
changes over the last 20 years, growing increasingly open to
trade since European Union (EU) integration in 1986. This has
affected domestic production processes, the technologies used,
and consumption patterns, which in turn influence water
demand and impacts. These developments have affected not
only direct but also indirect demand for water (via inputs).
Moreover, structural change has occurred in parallel with an
increase in trade flows with other EU member States and the
rest of the world (RW), which also affect water demand and
pressure on Spanish and foreign water resources. This occurs
because water consumption is driven by different factors
including production structure, trade composition, technology
and consumption habits as determinants of the water flows
within the economy. This study seeks to advance our
knowledge of water flows and the factors underlying them.
Specifically, this paper has a 2-fold objective: first, to evaluate
water flows in the different sectors in the Spanish economy and
interactions with the rest of the world; and second to use these
results for evaluating the pressure on water in different
spending scenarios for changes in Spanish diets. The key
concept in our analysis is the embodied water in a product, that
is to say, the water that directly or indirectly has been necessary
to produce it, but the concept of virtual water1 (VW) is also
used in parallel to show better the importance of the
now on, we will use as equivalent the concepts of VW content
of a product and embodied water in the product. Moreover, the
model used shows the water embodied in products bought by
domestic consumers, which we identify with the water
footprint2 (WF), and in trade, identified with VW trade.
One of the approaches to estimate the water embodied in
sectors/activities/products is input-output (IO) analysis. The
method applied is an open Leontief model based on the data
we have obtained for an extended social accounting matrix
(SAM) with water accounts, which allows us to estimate the
embodied water required to produce the goods and services
assigned to the exogenous accounts, in this study comprising
“households” and “foreign trade”. A number of papers illustrate
the possibility of using the VW concept in an IO framework to
analyze the water pressure derived from economic activities.3−5
In many of them, boundary input is given by water withdrawal
of the sector (here we will consider physical consumption) and
output, given by the water footprint (the sum of total domestic
water and the net virtual water import of the region, see more
on this in refs 6,7).
Received: March 1, 2011
Revised: May 7, 2012
Accepted: May 21, 2012
Published: May 21, 2012

© 2012 American Chemical Society 6530 dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538
Environmental Science & Technology
This methodology has been used at the national level for
Spain and for some regions,8,9 but not based on a disaggregated
table embracing the whole Spanish agri-food complex (defined
as the set of agricultural and livestock activities, the food
industry, and trade in all those products), given that the
country’s farm (i.e., agrarian) sector and food industry
represent an important share of national output, employment
and income generation. The use of a disaggregated table allows
precise estimates of the embodied water per euro in a product
of domestic final demand, or embodied in any product
imported or exported. Disaggregation of IO tables in activities
related with the agri-food complex is described in various
papers in the literature,10,11 and others deal with disaggregation
for the purposes of environmental analysis.12
This paper follows the IO methodology, offering a step
forward in the analysis of water uses, flows, trade and
consumption patterns in line with some of the new concepts
developed in the field of water management. We will focus on
physical consumption (PC, also called consumptive water use),
considering both the blue water collected from groundwater or
surface sources (the water usually studied in relation to the
political and economic management of the resource) and green
water (i.e., the water evaporated from rainwater stored in the
ground as soil moisture).
Section 2 describes the linear model and the items obtained
from the SAM. Section 3 contains the results and discussion of
the embodied water in the products of the Spanish economy,
which are completed with the valuations of embodied water in
imports and exports. Finally, we provide an analysis of how uses
might vary given shifts in domestic final demand, especially in
food consumption patterns.
2. DATA AND METHODOLOGY
Social Accounting Matrix for 2004. Our starting points
are the supply and use tables provided by the Spanish Statistics
Institute (NSI) and the Water Satellite Accounts (WSA) for
2004. Given our objectives and the availability of information
⎡ xP ⎤ ⎡ APP 0 APC APG APS ⎤⎡ xP ⎤ ⎡ yPH ⎤
⎢x ⎥ ⎢ ⎥
⎢ AFP 0 0 0 0 ⎥⎢ xF ⎥ ⎢ yFH ⎥
⎢ F⎥ ⎢ ⎥
⎢ ⎥
⎢ x C ⎥=
⎢ 0 A CF A CC A CG 0 ⎥⎢ x C ⎥ + ⎢ yCH ⎥ ⇔ x = Ax + y ⇔ x = (I − A)−1y
⎢x ⎥ ⎢ ⎥ ⎢
⎢ G⎥ ⎢A A GF A GC A GG A GS ⎥⎢ x G ⎥ ⎢ yGH ⎥
⎢ GP ⎥⎢ x ⎥ ⎢
⎣⎢ xS ⎦⎥ ⎢⎣ 0 0 A SC A SG 0 ⎥⎦⎣ S ⎦ ⎣ ySH ⎦
The accounts (block accounts) are P, productive sectors; F,
factors; C, companies; G, government; S, savings-investments;
and H, households and exports. With this choice of exogenous
accounts I−O, we place the focus on the pressure on the water
resources due to the final demand of goods outside the country,
and due to the final demand of the households (and easily per
citizen), although obtaining embodied coefficients with more
dispersion than traditional only sectoral I−O, and gathering
longer chains (a few interesting ones, such as capturing
embodied water not only due to the demand of goods and
services, but government expenditure or investments, which
lead to further demand of goods and services, etc.).
Blue direct PC of industrial and services activities is obtained
from the WSA provided by the NSI, and the agrarian direct
(blue and green) water from,23 which follows the methodology
Article
on both domestic water use and water uses associated with
trade, a Symmetric IO table including water accounts was built
for 2004 (at basic prices) using the industry technology
assumption.13
The SAM extends the information contained in the IO tables
to represent the circular flow of income. In this sense, although
the underlying models in IO and SAM are formally similar, all
the accounts involved to obtain a unit of income are considered
in the SAM models. The SAM used for the analysis is also
squared and allows the definition of linear models which are
formally analogous to IO linear models. So although we use the
information from a SAM, since we leave as exogenous accounts
the household consumption and foreign trade; it is not a
traditional SAM analysis, but more an extended input-output
analysis.
Having obtained the SAM table and extended it with water
accounts, we break down the agri-food system and the accounts
where water flows are expected to be most important, obtaining
a disaggregated extended SAM with water accounts for 2004
and 68 productive activities (2004 DSW), which is available in
the Supporting Information (SI). The water information was
also broken down for these accounts. For these purposes, our
main references are 14−21. Finally, a KRAS algorithm22 is
applied at different stages of the process in order to balance the
matrix, while preserving the values of some cells and groups of
cells, and to detect and correct conflicts in the data.
Modeling and Foreign Trade. The linear model used is
based on the DSW, taking households and foreign trade
(comprising the accounts of “European Union”, and “Rest of
the World”) as the exogenous. A matrix A of accounting
multipliers (which are formally similar to the technical
coefficients of I−O models) is obtained by dividing the
columns of endogenous accounts by the account total. The
following linear model, described in five blocks of accounts,
calculates the levels of output (resources), where x is the vector
of outputs for the endogenous accounts and y the exogenous
demand or final demand of the model:
⎢ ⎥
⎥
⎥
(1)
proposed, for example, in ref 17, but making a clear distinction
between dryland and irrigated land requirements. Aggregating
all types of water (we do not consider here gray water) we get
blue and green PC.
If we define a vector w of direct unit coefficients for water,
where {wi}i=1,...,n is the PC of each activity i per € value of total
output x of activity i, we can obtain a vector of water values by
category of final demand:
Λ′ = (λi)′ = w′(I − A)−1 (2)
These values capture all the water incorporated, directly or
indirectly, by unit of exogenous demand, so that λi is the
embodied water in each monetary unit of the Households and
Foreign trade accounts. The same formulation applies to any
other water vector w.
6531 dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538
Environmental Science & Technology Article

Table 1. Values of the Main Water Accounts Obtained from the 2004 DSW (hm3)
direct consumption
+ imports embodied water consumption in final demand
total total households exports
blue blue blue blue
GCa classifications−sectors blue +green blue +green blue +green blue +green
1 cereals and leguminous plants 6172 12 410 876 3048 613 2137 263 911
2 vegetables and fruits 4329 6956 3782 7864 1558 3252 2224 4612
3 industrial crops and woody 3510 4774 43 144 0 0 43 144
4−5 olive tree and grape wine 1634 4386 186 506 176 482 10 24
6−9 livestock 234 234* 248 1006 163 666 85 340
10−11 other agrarian, fishing and forestry 812 1301 105 398 74 302 31 96
activity
total agricultural, forestry and 16 699 30 070 5240 12 967 2584 6839 2656 6128
fishing (1−11)

12−17 extraction minerals and energy and 672 672 276 1446 211 849 67 597
water supply
18 chemicals 5 5 68 1158 15 252 53 906
19−22 metallurgy, machinery and equip. 93 93 186 4131 39 957 148 3174
23 meat industry 18 18 1675 7461 1388 6188 287 1273
24 diaries 8 8 619 3064 549 2717 70 347
25 industrial oils and greases 8 8 1178 4958 607 2554 571 2404
26−33 other food industries 36 36 2973 12 354 2134 8858 838 3496
34−38 beverages and tobacco 18 18 456 1875 283 1365 173 510
total food industry (23−38) 90 90 6901 29 712 4960 21 682 1941 8030
39−40 textiles, leather and footwear 25 25 338 3815 204 2312 133 1503
41−45 all other industries 170 170 236 1535 108 694 126 841

total industry (12−45) 1054 1054 8006 41 801 5538 26 745 2468 15 056
46−53 wholesale and retail agrarian and 2 2 600 2553 550 2342 51 211
food
54 nonfood trade 7 7 191 1120 163 929 35 191
55−56 restaurants, cafe, bars and similar 41 41 1963 7848 1692 6718 271 1130
57 −58 hotels, lodging and catering services 9 9 179 745 65 287 114 458
59−65 transport and private and domestic 31 31 656 3734 492 2763 164 971
services
66−68 public education, health and services 45 45 4 21 4 21 0 0
69−71 factors 0 0 0 0 0 0 0 0
72−73 NPISH and societies 0 0 143 882 92 565 50 317
74 public administration 0 0 539 3065 522 2962 17 103
75 savings/investment 0 0 367 2381 367 2381 0 0
76 households 201 201 201 810i 201 810b 0 0

1−76 total national (1−76) 18 090 31 461 18 090 12 266 5824

77 european union 14 623
78 rest of the world 31 840

77−78 total imports (77−78) 46 463

1−78 total (sum, 1−78) 77 924 0 77 924 0 53 364 0 24 560
a
GC: group code. bThe embodied imported water due to Spanish tourists’ consumption abroad is recognized in these accounts. *No direct green
(pastures...) water consumption considered.

To calculate the total PC of water associated with economic
activity, we must deal with three components: The water
consumed domestically to produce the goods and services
demanded by households and foreign trade, direct consump-
tion of water in exogenous activities, and embodied water in
imports (which we associate to VW import). Beginning with
the first component, if wdom=(wdomi ) is the vector of domestic
PC coefficients (defined accordingly with the total of the
matrix), the total water consumed in Spain to meet exogenous
demand yr (r = households, foreign sector) will be given by:
6532
∑ (widom)′(I − A)−1yr = ∑ wdom′(I − A)−1yr
r r (3)
The second component is associated with the direct
consumption of water in the activities or accounts that we
considered exogenous. If dEx r is the vector of water PC per
(monetary) unit of exogenous account r, this will be Σr dEx r ′yr.
To address the embodied water in imports, the third
component, we look at the technological conditions of the main
source countries for the (global and agrarian) products entering
the country as imports. Thus we have used IO tables24 for the
dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538
Environmental Science & Technology
countries from which Spain bought most imports in 2004
(tables harmonized in Euros), estimating more accurately the
embodied water of imports, based on the transformation
processes of the countries of origin, as improvement from the
single region approach. Table SI2 in the SI shows results for the
10 countries from which Spain imported most in 2004, which
were also the main sources of imports over the whole of the
decade, eight of them also being the main destinations of
exports. The countries concerned are France, Germany,
Portugal, Italy, UK, Netherlands, U.S., Belgium, China, and
Japan. However, other countries such as Argentina, Brazil,
Canada, South Africa, and Indonesia are also significant and
their IO tables are considered. Trade data were obtained from
the Spanish Revenue Agency.
This information was combined with water data for the
countries concerned to calculate vectors of water values for
each individual country and for groups of countries, c, as
follows::
Λcimp′ = (λciimp) = wc′(I − A c)−1 (4)
where wc is the vector of water consumption per unit of basic
output in c (and hence for every unit of imported output from c
to Spain), and Ac is the matrix of technical coefficients of c.
Once Λimpc and the vector of imports Mc=(Mci) are known, we
can obtain the unit coefficients for each country c as wimp c =
(wimp imp
ci ) = (λci Mci/xi) where xi is the total Spanish output of
sector i. Thus, the embodied water in Spanish production is
∑ w dom′(I − A)−1yr + ∑ drExyr
r r
+ ∑∑ w cimp′(I − A)−1yr
r c (5)
If we focus only on exports, assuming y for households to be
equal to 0, and that yr is export demand, the above expression
tells us that the embodied water in exports is
W exp = w dom′(I − A)−1yr +
∑ wcimp′(I − A)−1yr
c (6)
which comprises the domestic water consumed to obtain
exports, plus the embodied water in inputs imported to
produce them.
Uncertainty in the Model. Finally, we consider the
underlying uncertainties following this long process of
combining tables, matching and balancing data sources,
comprising those inherent in the primary input-output, trade
and water data, the detailed data used for the disaggregation,
price indexes, and model uncertainties.25 A summary of the
sensitivity analysis performed in this respect is shown in the
results. The approach used is not new in the field (for example,
see refs 26 and 27) but relatively little attention has been paid
to this issue, becoming recently a matter of concern in
multiregional IO models, environmental IO models,28−30 and
with respect to the WF.31
3. RESULTS AND DISCUSSION
3.1. Blue and Green Water Account for the 2004 DSW.
Estimates for blue and green water associated with both
production and consumption, appear in Table 1. The first and
third columns of numerical results respectively show direct and
embodied blue consumption, given the cost of the blue
resource and its importance for many social and economic
activities. The total (blue and green water) direct and
6533
Article
embodied consumption is shown in the second and fourth
columns. The break down between households and exports for
both types of water can be seen in columns 5 to 8. The pressure
of the Spanish households (citizens) on the global water
resources (which we associate with our boundary output, the
blue and green WF), represented in the sixth column in Table
1, comes from the domestic water consumptive use (WC), plus
the embodied water in imports (VWimp), see second column,
minus embodied water in exports (VWexp): WF = WC + VWimp
− VWexp
The total PC of water in production plus embodied water in
imports and direct uses, as expressed in eq 5 (which we relate
to WF budget), is 77 924 hm3/year, while the embodied water
in imports (third addend of eq 5) is 46 463 hm3/year, the 60%
of the total PC. From the total PC only 18 090 hm3/year is
domestic blue water (less than one-fourth), being consumed
most of them, 16 699 hm3/year, in the agrarian activities and
the rest in the industrial production and in services. Also to
agriculture corresponds the highest domestic green water
consumption, 13 371 hm3/year.
In Table 1 we may see that more than 60% of the imported
water comes from non-EU countries, and as we will see, mainly
from primary and transformed farm products. Similar results
were obtained in ref 17 As shown in the last column of Table 1
(eq 6), meanwhile, the total embodied water in exports is 24
560 hm3/year, about a third of the total PC. Hence, Spain is a
net importer of VW, in line with previous process analysis type
of studies (as defined in ref 32), focused on agrarian and food
trade, as seen in refs 17,23
As may be seen in Table 1, the embodied water volumes in
the final demand from Food industry (29 712 hm3) are more
than 300 times the direct consumption of water of the sector,
demonstrating the relevance of water flows from other
activities, especially Agriculture and Livestock. A similar result
is observable for the total of Industries (41 801 hm3) which is
more than 39 times its direct consumption of water, due mainly
to differences in the Food industry, and in industrial accounts,
as textiles, clothing, leather, and footwear.
Flow transfers in the food industry can be better understood
by focusing on the different sectors involved. The meat industry
directly consumes only 18 hm3 of water, but the water needed
to satisfy final demands (households or exports) is 7,461 hm3;
other subindustries such as dairies, and industrial oils and
greases directly consume about 8 hm3, but their embodied
consumption is 3064 hm3 and 4958 hm3 respectively.
Analogously, the restaurants, cafe, bars and similar directly
consume 41 hm3, but the water needed to satisfy final demands
is high at around 7848 hm3.
3.2. Water Consumption and Spanish Trade. According
to the results presented above, Spain is a net importer of water.
As much of Spain is arid or semiarid, it will be very interesting
to examine the VW exported and imported in greater depth.
Taking only domestic PC into account, we may observe that
12 266 hm3 of the total 18 090 hm3 of domestic blue water
consumed every year remain in the country, while the
remaining 5823 hm3 of water, almost 1/3 of the total, is
exported. We find a similar picture for total consumption (blue
and green water,): 53 364 hm3 of the total 77 924 hm3 (68.5%)
are embodied in household consumption, whereas the
remaining 24 560 hm3 (31.5%) is exported. The most
significant sectors in terms of blue VW exports are vegetables
and fruits; pork products; chemicals; transport equipment;
industry of vegetables, of milling, of wines and ciders, rubber,
dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538
Environmental Science & Technology Article

Scheme 1. Schematic Main Net Water Flows (Blue -Big Flows- and Black Arrows) Across Spanish Sectors (Orange Ovals) and
with Other Countries (Red Arrows)a,b

a
BE: Belgium, GE: Germany, FR: France, GB: Great Britain, IT: Italy, NL: The Netherlands, PT: Portugal, Other EU: Other European Union
countries, JP: Japan, US: United States, CH: China, BR: Brazil, AR: Argentina, RUS: Russia, CAN: Canada, IN: India, INDO: Indonesia, TUR:
Turkey, SA: South Africa, SK: South Korea, RW: Rest of the World. bBlue arrows show the main water flows across sectors of the Spanish economy
(ovals), being the black ones less important in terms of water. The size of the selected main net flows is given by the size of the spheres represented,
while the importance of the agrarian sector in the flow is given by the color, being the yellow ones those with the highest water content from the
Agrarian Sector.

plastics, and other manufactures; hotel, pensions and similar;
and transport and communications.
Supplementing the above information, the map in Scheme 1
(and Table SI2 in the SI) present the main net water flows
across sectors and embodied in net exports.
Spain is a net importer of embodied water flows, especially
from non EU countries, but generally is net exporter to EU
partners, 18 986 hm3 versus 14 623 hm3, although clearly not
with France, from where much water comes from agrarian and
forestry goods. Spanish VW exports to the United Kingdom,
Italy and Portugal are also fairly significant, especially in the
case of agri-food products (standing out Vegetables and Fruits,
especially citrus). Meanwhile, embodied water in imports from
the U.S. (2467 hm3), China (3115 hm3), and the RW (25,749
hm3), see SI Table SI2, may avoid considerable domestic
consumption, as the water balance with these countries (with
importance of the agrarian sector in the VW flows, as given by
the yellow spheres in each country) is significantly negative.
Notably, around 50% of the embodied water in imports from
China comes from textile products, leather, and footwear.
As discussed in the SI in the subsection of uncertainty in the
scenario analysis, scenario analysis about trade would benefit
from a model where trade is endogenous.
3.3. Scenarios Referring to Changes in Diet and
Household Water Consumption. One of the advantages of
working in an IO framework, with disaggregated sector
information, is that it allows simulation of different scenarios
6534
for changes in economic behavior and appraisal of their effects
on economic and environmental flows. This section is a first
approximation to the changes in the water consumed in the
face of variations in the food products purchased by
households, since we have identified that expenditure as one
of the key determining the WF. Moreover, according to various
references,33,34 consumption of meat and meat products in the
Spanish diet is higher than recommended, but less cereals and
cereal products, vegetables, fruit, and legumes are eaten than
would be desirable.
In this context, in Table 2 we present two scenarios based on
a comparison of actual and recommended diets. In Scenario I,
changes in spending are associated with a recommended diet
by36 for Spain. Changes of this nature would reduce spending
on food by 11 611 Euros, 1.51% of household expenditure. In
Scenario II, the changes in agriculture, livestock and food
spending are accompanied by the return of the savings
generated to the economic circuit, in the form of higher
spending on other activities. This approximates with a simple
assumption a possible rebound effect from the extra money
generated with Scenario I. The necessary data on food supply
quantities (in kg per capita per year, by crop and by specific
animal classification) is found in ref 35 for 2004. The
percentage change in spending is estimated from the ratio of
changes between the recommended diet and the current diet,
assuming no changes in prices.
dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538
 Environmental Science & Technology
Table 2. Variations in Embodied Water (hm3) in the Main Accounts from Dietary Changes
current situation scenario i scenario ii
household embodied % of household embodied % of variation household embodied % of
expendit. consumption embodied variation in expendit. consumpt. of embodied in expendit. consumpt. of embodied
a b
GC classifications−sectors (million €) of water consumpt. expendit. (million €) water consumpt. expendit. (million €) water consumpt.
1 cereals & leguminous plants 708 2137 4.05% 8% 764 2306 4.82% 8% 764 2306 4.79%
2 vegetables & fruits 5042 3252 6.16% 21% 6119 3946 8.25% 21% 6119 3946 8.20%
3 industrial crops 0 0 0% 0% 0 0 0% 0% 0 0 0%
4−5 olive trees & vineyards 449 482 0.91% 0% 449 482 1.01% 0% 449 482 1.00%
6−9 livestock 613 666 1.26% −82% 112 121 0.25% −82% 112 121 0.25%
10−11 other agrarian, fishing & forestry 2321 302 0.57% −39% 1465 225 0.47% −39% 1465 225 0.47%
1−11 total agrarian s., forestry & fish. 9134 6839 12.96% −2% 8909 7080 14.80% 1.86% 8909 7080 14.71%

12−17 extraction minerals, energy & water 14 897 849 1.61% 0% 14 897 850 1.78% 0% 15 174 866 1.80%
supply
18 chemicals 4589 252 0.48% 0% 4589 252 0.53% 1.86% 4674 257 0.53%
19−22 metallurgy, machinery & equip. 21 525 958 1.82% 0% 21 525 958 2.00% 1.86% 21 927 975 2.03%
23 meat industry 11 785 6188 11.73% −82% 2149 1128 2.36% −82% 2149 1128 2.34%
24 dairies 6429 2717 5.15% −1% 6345 2681 5.60% −1% 6345 2681 5.57%
25 industrial oils & fats 2008 2554 4.84% 0% 2008 2554 5.34% 0% 2008 2554 5.31%
26−33 other food industries 15 798 8857 16.79% −10% 15 272 8993 18.79% 0% 15 272 8993 18.69%
34−38 beverages & tobacco 5688 1366 2.59% 0% 5688 1366 2.85% 0% 5688 1366 2.84%
6535

23−38 total food industry (23−38) 41 709 21 682 41.10% −25% 31 463 16 721 34.94% −25% 31 463 16 721 34.74%

39−40 textiles, leather & footwear 13 183 2312 4.38% 0% 13 183 2312 4.83% 1.86% 13 429 2355 4.89%
41−45 all other industries 14 251 694 1.32% 0% 14 251 694 1.45% 1.86% 14 518 707 1.47%

12−45 total industry (12−45) 110 153 26 747 50.70% −1% 99 907 21 788 45.53% 1.86% 101 184 21 882 45.47%
46−53 wholesale agrarian & food 4.48% 26 826 2342 4.44% 25 686 2154 4.50% 0% 25 686 2155
dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538

−100%
54 nonfood trade 49 748 929 1.76% 0% 49 748 929 1.94% 1.86% 50 675 946 1.97%
55−56 restaurants, cafe, bars & similar 69 231 6718 12.73% 0% 69 231 6718 14.04% 0% 69 231 6718 13.96%
57−58 hotels, lodging & catering services 5351 287 0.54% 0% 5351 287 0.60% 1.86% 5451 293 0.61%
59−65 transport & private & domestic 157 241 2763 5.24% 0% 157 241 2763 5.77% 1.86% 160 171 2814 5.85%
serv.
66−68 public education, health & services 2084 22 0.04% 0% 2084 22 0.05% 1.86% 2123 22 0.05%
69−71 factors 0 0 0.% 0% 0 0 0% 0% 0 0 0%
72−73 NPISH & companies 31 623 566 1.07% 0% 31 623 566 1.18% 1.86% 32 213 576 1.20%
74 public administration 213 694 2962 5.61% 0% 213 694 2962 6.19% 1.86% 217 677 3018 6.27%
75 savings/investment 61 663 2381 4.51% 0% 61 663 2381 4.98% 1.86% 62 812 2426 5.04%
76 households 21 235 201c 0.38% 0% 21 235 201 i 0.42% 1.86% 21 630 201 i 0.42%
77 european union 6944 0 0% 0% 6944 0 0% 1.86% 7074 0 0%
78 rest of world 5013 0 0% 0% 5013 0 0% 1.86% 5107 0 0%
1−78 total all accounts (1−78) 769 943 52 756 100% −1.51% 758 332 47 850 100% 0% 769 943 48 129 100%

Article
a
GC: group code. bserv.: services; NPISH is non profit institutions serving households. cThe embodied imported water due to Spanish tourists’ consumption abroad is not recognized here (hence resulting
in a different total than in Table 1).
Environmental Science & Technology
Scenario I results in a 21% increase in both fresh and
prepared vegetables and fruit, an 8% increase in cereals and
legumes, an 82% decrease in livestock and meat products and,
finally, a 39% decrease in both fresh and canned fish. The water
saving associated with the recommended diet is relevant, 4906
hm3, the 9.30% of total, mainly because of the large increase in
the consumption of cereals and legumes, and vegetables and
fruit, and the decrease in meat consumption. The significant
change in meat consumption outweighs the switch to cereals,
legumes and fruit and vegetables even when the water content
of meat per euro of expenditure (VW intensity) is lower (as it is
also a more expensive product).
While maintaining the changes in the food accounts
described above, Scenario II examines the change in the total
WF when the extra money (11 611 Euros) obtained from lower
expenditure on food products is spent on other activities not
directly related with food consumption. Current expenditure on
these activities is 623 043 million Euros, so this would represent
an increase of around 1.86% with respect to current spending
on these goods. The resulting reduction in the WF is smaller
than in Scenario I, although it is still high, 8.77%. Looking at
the participation of the different sectors under the baseline
situation and scenarios, columns third, seventh and 11th in
Table 2, we may observe that water reductions are mainly
driven in both Scenario I and II by reductions in embodied
water consumed from livestock and meat, which is one of the
most representative in water consumption. As it can be seen,
scenarios affect not only the water embodied in the different
sectors but also the fraction represented by each of them (%
with respect to total consumption). Thus, embodied water
becomes higher in the simulated scenarios than in the current
situation for sectors such as industrial oils and fats, wholesale
agrarian and food, restaurants, cafe, bars and similar or diaries
and other food industries (even when from this two last ones it
is reduced the absolute embodied water consumption). In any
case, the results indicate that a move toward healthier diets
would lead to a lower WF.
A change in diets also implies altering the proportion of
domestic to imported water that it is finally consumed by the
Spanish households. In the base scenario, 40% of water PC
necessary to satisfy Spanish households’ demands is domestic
water, being 60% the imported water. With the change in diets,
assuming the current structure of production (see subsection
Exogenous bilateral and total trade f lows in the SI for the analysis
of further changes in the structure of production), it should be
expected a reduction on both domestic and VW import, being
bigger the reduction in VW import. That reduction would be
especially marked in those from Italy, Poland, Denmark,
Sweden, etc. (high relative imports from livestock); on the
contrary, due to increase in beans and leguminous con-
sumption, VW imports from U.S., UK, Canada, Mexico, and
France, would not vary that much.
Finally regarding the water stress, some tools and date used
can also be applied to obtain country-based calculations and
additional interpretations of the water footprints estimated (see
ref 36). To begin with differential characteristics, we may note
the significant share of agricultural water use in Spain and the
high ratio of total annual freshwater withdrawals and
consumption to availability (leading to high “stress indexes”)
compared to European and other countries at a similar stage of
development. Also, consumptive uses exact a high toll in terms
of adverse impacts on ecosystem quality. Given the limits and
differences in approach (our focus was global and did not
6536
Article
distinguish between hydrological basins), a first approximation
reveals that a 10% cut in domestic water consumption
(assumed to be equal for the Northwest, Northeast, and
Central, and South) would shift the ratio of water consumption
to hydrological availability from approximately 20%, 40%, and
80% for each of the three regions to 15%, 30%, and 70%.
3.4. Uncertainty and Sensitivity of the Results to Data
Sources and Modeling. As discussed more extensively in the
SI, we have used various different methods to check our data
and results. With respect to virtual (embodied) water content,
Figure 1 presents Monte Carlo simulations (1000 perturbations
Figure 1. Frequency distribution of deviations of perturbed water
multipliers M. As variation from their unperturbed value. The
distribution is obtained from 1000 perturbations of F* and A*,
resulting in more than 100 000 perturbed multipliers.
of F* and A*) of standard errors in VW multipliers (or
embodied intensities), showing that they are much smaller
(almost all below 10% and mostly below 5%, with a peak
around 0%) than in the coefficients matrix A* (ranging up to
60% as defined by the authors based on ref 32) and direct water
intensities F* (up to 150%). Finally, with respect to the current
food consumption, its information is highly detailed and the
main uncertainties are probably due to consideration of
scenarios that are too general in terms of the recommended
changes. Our main conclusion, based on these sensitivity
scenarios, is that although we should look at issues such as
possible changes in the proportion of blue and green, in trade
(especially with the scenarios of changes in diets) or in
consumption of food related industries, the most important
variations in the estimations would arise from agri-food system
data aggregation.
4. IMPLICATIONS AND FURTHER RESEARCH WITH
THE 2004 DSW
This paper offers a detailed study of the water flows associated
with economic activity in Spain. To this end, a SAM with a high
level of disaggregation in the agri-food system was built and
extended to include water accounts, unlike the usual IO models,
containing few agri-food accounts. This represents a step
forward with respect to usual IO model with few agri-food
accounts, and to other analyses with precise agri-food
information, but narrow system boundaries selection,37 by
trying to profit best the advantages of both approaches. The
paper thus implicitly highlights the importance of breaking
down the agri-food system in order to obtain relevant water
intensities. Also, since the imported and exported water
volumes are so high, although we did not construct a
dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538
Environmental Science & Technology
multiregional model, import coefficients were carefully
considered using IO tables, process analysis type of studies
as,17 water data, and data on agricultural imports from source
countries. The main characteristics of blue and green water
accounting in the Spanish economy were examined on the basis
of this information, which could easily be extended to further
studies of technological changes (e.g., in irrigation) and impacts
on the environment using the 2004 DSW.
The results indicate that Spain is a net importer of VW, being
more than 60% of VW imports sourced from countries outside
the EU (though France stands out as major source), and from
primary and transformed agrarian products.
The analysis of sensitivity to changes in diet suggests the
possibility of obtaining water use reductions through changes in
household behavior. The diet variation scenarios presented here
are relevant, given the high water intensity of food and because
making policy recommendations to households is likely to be
more effective than addressing agents (e.g., consumers) abroad.
Also, although an accurate description of the consequences on
water demand of different dietary scenarios would require
considering issues as elasticities, prices and substitution effects,,
the results highlight that there is room for water reductions
moving toward a healthier diet. The results confirm a 9%
reduction in the WF by following scientifically supported
recommendations (see refs 34 and 35) to eat a healthier diet
including less meat (which is avidly consumed in Spain).
We have revealed three important aspects regarding both
Spanish households’ consumption and water use in Spain. First,
the estimation of the WF has shown that the households are
responsible of an important part of total water consumption in
Spain, being this consumption highly linked to the
consumption patterns and in particularly to a diet with high
meat content. Thus, changes in consumption patterns may
contribute to alleviate water pressures in Spain. Second, we
have seen that a significant part of water consumption in Spain,
is to some extent, avoided through imports of products, being
Spain a net VW importer. The results show that trade policy
and the international context have an important effect on the
future of the Spanish water resources. Tools like the ones
presented serve us to identify unsustainable paths, and simulate
scenarios and policies in order to reverse those processes. We
believe that this type of disaggregated information in a SAM
framework enables interesting applications linking fiscal policies
and natural resources.
ASSOCIATED CONTENT
■
*
S Supporting Information
Additional information as noted in the text. This material is
available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*Phone +34 976-761826 and +34 646019737; fax: +34-976-
761996; e-mail: icazcarr@unizar.es.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank the anonymous reviewers for their helpful comments
in previous versions of the paper which improved the work
considerably. Ignacio Cazcarro also thanks the Spanish Ministry
6537
Article
of Science and Innovation for a FPU grant (AP2007-03816).
Any remaining errors are our sole responsibility.
■ REFERENCES
(1) Allan, J. A. Policy responses to the closure of water resources:
regional and global issues. In Water Policy: Allocation and Management
in Practice; Howsam, P.; Carter, R. C., Eds.; E&FN Spon: London,
1996; pp 3−12.
(2) Hoekstra, A. Y.; Hung, P. Q. Virtual water trade: A quantification
of virtual water flows between nations in relation to international crop
trade. Virtual Water Trade: A Quantification of Virtual Water Flows
Between Nations in Relation to International Crop Trade 2002.
(3) Lenzen, M. Victorian Water Trust. Report on the Virtual Water
Cycle of Victoria; GHD: Melbourne. Victoria, 2008.
(4) Yu, Y.; Hubacek, K.; Feng, K. S.; Guan, D. B. Assessing regional
and global water footprints for the UK. Ecol. Econ. 2010, 69 (5),
1140−1147.
(5) Zhao, X.; Yang, H.; Yang, Z.; Chen, B.; Qin, Y. Applying the
input-output method to account for water footprint and virtual water
trade in the haihe river basin in China. Environ. Sci. Technol. 2010, 44
(23), 9150−9156.
(6) Yang, Z. F.; Mao, X. F.; Zhao, X.; Chen, B. Ecological network
analysis on global virtual water trade. Environ. Sci. Technol. 2012, 46
(3), 1796−803.
(7) Mao, X. F.; Yang, Z. F. Functional assessment of associated
aquatic ecosystems of associated aquatic ecosystems in the
Baiyangdian BasinAn ecological-network-analysis based approach.
Ecol. Model. 2011, 222 (23−24), 811−3820.
(8) Velázquez, E. Water trade in Andalusia. Virtual water: An
alternative way to manage water use. Ecol. Econ. 2007, 63 (1), 201−
208.
(9) Duarte, R.; Sánchez-Chóliz, J.; Bielsa, J. Water use in the Spanish
economy: An input-output approach. Ecol. Econ. 2002, 43 (1), 71−85.
(10) Peterson, E. Disaggregating Agricultural and Food Sector (Chapter
15), GTAP 4 Data Base Documentation. Center for Global Trade
Analysis; Purdue University, 1999.
(11) Müller, M.; Pérez Domínguez, I; Gay, S. H. Construction of
social accounting matrices for the EU-27 with a disaggregated
agricultural sector; Institute for Prospective Technological Studies.
JRC European Comission, 2010.
(12) Lenzen, M. Aggregation versus disaggregation in input-output
analysis of the environment. Econ. Syst. Res. 2011, 23 (1), 73−89.
(13) Rueda-Cantuche, J. M.; Beutel, J.; Neuwahl, F.; Mongelli, I.;
Loeschel, A. A symmetric input-output table for EU27: Latest
progress. Econ. Syst. Res. 2009, 21 (1), 59−79.
(14) Titos Moreno, A.; de Haro, T.; Gómez Muñoz, A. C.; Ramos
Real, F. El Sistema Agroalimentario Español: Tabla Input-Output y
Análisis de las Relaciones Intersectoriales; Ministry of Agriculture,
Fisheries and Food in Spanish, General Technical Secretary: Madrid,
1995.
(15) FAO, Technical Conversion Factors for Agricultural Commod-
ities, Food and Agriculture Organization of the United Nations, Rome,
Italy, 2003.
(16) Chapagain, A. K.; Hoekstra, A. Y. Virtual Water Flows Between
Nations in Relation to Trade in Livestock and Livestock Products, Value of
Water Research Report Series 13; UNESCO-IHE, Delft, the
Netherlands, 2003.
(17) Chapagain, A. K.; Hoekstra, A. Y. Water Footprints of Nations,
Value of Water Research Report Series 16; UNESCO-IHE: Delft, the
Netherlands, 2004.
(18) MAPA. Agrarian Accountancy Net (RECAN); Ministry of
Agriculture, Fisheries and Food: Madrid, Spain, 2004 (in Spanish).
(19) National Statistics Institute (NSI). Accounting Series (Spanish
National Accounts, Industrial Products Survey, Annual Survey of Services,
Annual Survey of Trade, Continuous Household Budget Survey, Wage
Structure Survey, Labour Force Survey, Health, Tourism Satellite Account,
Satellite Accounts of water, Survey of Water Supply and Treatment,
Surveys on Water Use in Agriculture and in Industry); National Accounts
of Spain: Madrid.
dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538
Environmental Science & Technology Article

(20) FAO. Aquastat. Food and Agriculture Organization of the

United Nations: Rome, Italy, 2007.
(21) Eurostat, Environmentl-Water- Annual water abstraction by
source and by sector, 2010.
(22) Lenzen, M.; Gallego, B.; Wood, R. Matrix balancing under
conflicting information. Econ. Sys. Res. 2009, 21 (1), 23−44.
(23) Garrido, A.; Llamas, M. R.; Varela-Ortega, C.; Novo, P.;
Rodríguez-Casado, R.; Aldaya, M. M. Water Footprint and Virtual
Water Trade in Spain: Policy Implications; Springer - Fundación
Marcelino Botin,́ 2010.
(24) Yamano, N.; Ahmad, N. The OECD Input Output Database:
2006 ed.. 2006/8; Organization for Economic Cooperation and
Development: Paris, 2009.
(25) Weber, C. L. Uncertainties in constructing environmental
multiregional input-output models. Paper presented at the Interna-
tional input-output meeting on managing the environment, 2008,
Seville, Spain.
(26) Bullard, C.; Sebald, A. Monte Carlo sensitivity analysis of input-
output models. Rev. Ec. Stat. 1988, 70, 708−712.
(27) Lenzen, M.; Pade, L.; Munksgaard, J. CO2 multipliers in multi-
region input-output models. Econ. Syst. Res. 2004, 16 (4), 391−412.
(28) Wiedmann, T.; Lenzen, M.; Wood, R. Uncertainty Analysis of the
UK-MRIO Model − Results from a Monte-Carlo Analysis of the UK
Multi-Region Input-Output Model (Embedded Carbon Dioxide Emissions
Indicator), Report to the UK Department for Environment, Food and
Rural Affairs by Stockholm Environment Institute at the University of
York and Centre for Integrated Sustainability Analysis at the
University of Sydney; Defra: London, UK, 2008.
(29) Weber, C. L.; Matthews, H. S. Embodied emissions in U.S.
international trade: 1997−2004. Environ. Sci. Technol. 2007, 41 (14),
4875−4881.
(30) Yamakawa, A.; Peters, G. P. Using time-series to measure
uncertainty in environmental input-output analysis. Econ. Sys. Res.
2009, 21 (4), 337−362.
(31) Hoekstra, A. Y.; Chapagain, A. K.; Aldaya, M. M.; Mekonnen,
M. M. The Water Footprint Assessment Manual: Setting the Global
Standard, Earthscan: London, UK, 2011.
(32) Lenzen, M. A generalized input-output multiplier calculus for
Australia. Econ. Syst. Res. 2001, 13 (1), 65−92.
(33) SENC. Guiá de la alimentación saludable. Sociedad Española de
Nutrición Comunitaria, 2004, 2007.
(34) Varela Moreiras, G.; del Pozo, S.; Á vila, J. M.; Cuadrado, C.;
Ruiz, E.; Moreiras, O. Valoración de la dieta Española De Acuerdo al
Panel de Consumo Alimentario; Ministerio de Medio Ambiente, y
Medio Rural y Marino/Fundación Española de la Nutrición, 2008.
(35) FAO. Food Balance Sheets/ Commodity Balances: Crops Primary
Equivalents and Livestock and Fish Primary Equivalents. Agriculture and
Food Trade: Crops and Livestock, Primary and Processed; Food and
Agriculture Organization of the United Nations: Rome, Italy, 2010.
(36) Pfister, S.; Koehler, A.; Hellweg, S. Assessing the environmental
impacts of freshwater consumption in LCA. Environ. Sci. Technol.
2009, 43 (11), 4098−4104.
(37) Suh, S.; Lenzen, M.; Treloar, G. J.; Hondo, H.; Horvath, A.;
Huppes, G.; Jolliet, O.; Klann, U.; Krewitt, W.; Moriguchi, Y.;
Munksgaard, J.; Norris, G. System boundary selection in life-cycle
inventories using hybrid approaches. Environ. Sci. Technol. 2004, 38
(3), 657−664.

6538 dx.doi.org/10.1021/es203772v | Environ. Sci. Technol. 2012, 46, 6530−6538