Virtual Water Trade in the Mediterranean: Today and Tomorrow

Abstract

Virtual water trade refers to the implicit content of water in the production of goods and services. When trade is undertaken, there is an implicit exchange of water. In countries where water is relatively scarce, water-intensive goods are expensive to produce, so that imports normally exceed exports and the economy virtually imports water. This paper provides some estimates of virtual water trade patterns in the Mediterranean, which is an area where water is scarce, unevenly distributed, and progressively insufficient because of climate change and reduced precipitation. We analyse two cases: the current virtual water trade structure, related to trade in agricultural goods, and a future scenario, simulated by means of a computable general equilibrium model, where reduced agricultural productivity, induced by lower water availability, is taken into account.

1 Introduction

Water availability is a key factor in many societies, shaping cultures, economies, history, and national identity. This is especially true in the Mediterranean, where water resources are limited and very unevenly distributed over space and time.

There is a growing concern about water resources in this region. On the demand side, during the second half of the twentieth century, water demand has increased twofold, reaching 280 km³/year (UNEP 2006). Much of the demand comes from agricultural activities (45 % in the North, 82 % in South and East), but other industries also contribute significantly (most notably, tourism) and more competition for water resources can be easily foreseen in the near future.

Availability of water resources affects international trade and the relative competitiveness of countries and industries. In this context, “virtual water” is a useful concept to highlight the link between water consumption and trade (Allan 1993). The virtual water content of a good is defined as the volume of water that is actually used to produce that product. This will depend on the production conditions, including place and time of production and water-use efficiency. Producing one kilogram of grain in an arid country, for instance, can require two or three times more water than producing the same amount in a humid country (Hoekstra 2003).

When a good is exported, its virtual water content is implicitly exported as well. Vice versa, when one good is imported, the water used in its origin country of production is virtually imported. An origin/destination trade matrix can therefore be translated in terms of virtual water equivalent flows, allowing one to see whether a country is a net importer or exporter of virtual water, and which are its trade partners.

A large and flourishing literature on virtual water, as well as on the related concept of water “footprint”, is now available (for a critical review, see Yang and Zehnder 2007). Recently, the National Geographic (2010) magazine provided a map of virtual water trade flows in the world. The idea behind the virtual water concept is not restricted to water, but applies equally well to any resource, for example carbon, so we can also discuss about “virtual carbon” trade (Atkinson et al. 2010), that is, carbon emissions generated by foreign consumption (more often named “carbon leakage”).

In this paper, we estimate and analyse the current virtual water trade flows for some countries in the Mediterranean. Following Antonelli et al. (2012), we consider both the direct and the indirect (that is, associated with intermediate factors) water consumption, and we make a distinction between “green” and “blue” water.

We compare the current picture of virtual water trade in the Mediterranean with a counterfactual one, where we simulate the effects of reduced water availability. To this end, we first construct a scenario, accounting for changes in water demand and supply at the year 2050. We then estimate the associated changes in productivity for agricultural industries in some Mediterranean countries. Subsequently, we simulate the effects of changing productivity on international trade by means of a Computable General Equilibrium model of the world economy (Hertel and Tsigas 1997). The counterfactual trade patterns estimated by the CGE model are then translated in terms of virtual water trade flows, so that an assessment can be made about how reduced water availability would affect the structure of virtual water trade in the Mediterranean.

The paper is organized as follows. In the next section, some estimates of current virtual water trade flows in the Mediterranean are presented and discussed. Section 3 illustrates how future changes in water availability, and the related impact on agricultural productivity, have been obtained. Section 4 presents the results of the CGE simulation exercise in terms of virtual water trade. A final section provides some concluding remarks.

2 Current Virtual Water Trade in the Mediterranean

To estimate current virtual water trade flows, we consider 11 countries and 3 regional economies, obtained through aggregation from the GTAP 8 database.¹ These are Albania, Croatia, Cyprus, Egypt, France, Greece, Italy, Morocco, Spain, Tunisia, Turkey, Rest of Europe (Xeur), Rest of Middle East and North Africa (XMENA), and Rest of the World (RoW). Chapagain and Hoekstra (2004) provide estimates of total water consumption for 164 crops in 208 countries. We aggregate data to the 14 regions and 7 agricultural industries of the GTAP database, and then, we make a comparison between water consumption, by crop and region, and value of production (2004). This creates an estimate of direct water usage by unit of output (in monetary terms).

The direct water usage should not be confused with the unit virtual water content, as the latter includes the water indirectly consumed through the utilization of intermediate production factors. Antonelli et al. (2012) show how virtual water coefficients can be estimated, while taking into account the input–output linkages among sectors in the economy. We apply this methodology to get the systemic virtual water consumption, per unit of output, for each agricultural product in all regions. These parameters allow translating trade flows into equivalent virtual water units.

The whole matrix of bilateral virtual water trade flows corresponding to agricultural trade flows in the 2004 GTAP 8 database is displayed in the Appendix. In the following, we illustrate some summary indicators of current virtual water trade in the Mediterranean, to highlight some of its key characteristics.

Figure 1 shows the per capita virtual water trade balance, that is the difference between total virtual water exports and imports. It can be interpreted as a measure of trade-related water dependence. Countries where water resources are scarce and expensive are generally expected to have a comparative disadvantage in water-intensive industries, resulting in net virtual water imports. However, this outcome may not necessarily emerge, because of a number of other effects at work, like market failures (severe in the water sector, where resources may be underpriced and over-exploited) and other factors affecting the overall competitiveness of an industry in the world economy.

Fig. 1

Per capita net virtual water exports/balance (Mm³)

We can see that Cyprus is the country which indirectly obtains the largest amount of water through trade in agricultural goods, followed by Rest of Middle East and North Africa, Italy, and Greece.

More information about the sources of virtual water trade can be obtained through a decomposition of flows by region of origin and destination, which is provided in Fig. 2. In this figure, each country and region is associated with a negative and a positive bar. Negative bars express imports and positive bars express exports. Each bar is split in terms of countries of origin and destination, all with a distinctive colour. The trade balance displayed in Fig. 1 is just the algebraic difference between total exports and total imports in Fig. 2.

Fig. 2

Per capita virtual water trade flows in Mediterranean countries (Mm³)

Figure 2 highlights some interesting facts. First, some countries (e.g. Cyprus and XMENA) have a polarized trade: they import a lot but make almost no exports. Other countries, like Spain, are more balanced: they have quite large input and output flows, suggesting that they may play a role as hub in agricultural markets, possibly by importing raw agricultural goods and exporting refined, processed goods. Second, some countries have principal and different trading partners for imports and exports. Spain, but also Morocco and Italy, gets the bulk of their virtual water imports from outside Europe and the Mediterranean, exporting mainly towards central and northern Europe.

Figure 3 provides a picture of major virtual water flows on a geographical map. It displays only the largest flows, where thickness of the arrow line depends on the flow magnitude.²

Fig. 3

Largest flows of virtual water trade in the Mediterranean

The most significant exchanges of virtual water are found between the largest North-Mediterranean economies. France and Spain are the greatest traders of agricultural goods. An important role is also played by Italy, which is a substantial importer of agricultural products, and by some North African countries, such as Morocco and Tunisia, as well as by Turkey.

Further insights can be obtained by making a distinction between green and blue water trade flows. Green water is water stored into the soil moisture. Blue water is surface and underground water. The distinction is important because, whereas green water can be used only in agriculture, blue water can be allocated to alternative uses. Therefore, it has a greater economic potential and value. To the extent that virtual water trade is an indirect mean of saving on water resources, savings obtained on precious blue water is what matters the most.

From our estimates, it is possible to assess the blue virtual water “savings” achieved by means of trade in agricultural goods. This is done by computing how much blue water would be needed if imports were instead produced domestically, while subtracting from the latter the blue water consumed to produce the exported goods. Therefore, this new balance expresses how much extra blue water would have been necessary if a country would not have been involved in international trade, at given (unchanged) levels of domestic consumption. Results are shown in Fig. 4.

Fig. 4

Per capita blue water “savings” through trade in agricultural goods (Mm³)

Much of the countries “save” on blue water by trading agricultural goods. Albania and Cyprus display the most significant savings, followed by Rest of Middle East/North Africa, Egypt, Italy, and Greece. France is the only country exhibiting an exploitation of blue water resources, but this comes quite naturally, as blue water is relatively abundant in that country.

3 A Scenario of Future Water Availability for the Mediterranean

Virtual water patterns will change in the future because international trade will. Any factor affecting the world economy, including growth, economic policies, demography and external shocks, has the potential to generate repercussions on virtual water flows.

In this paper, we focus on the prospective effects of varying water availability in the Mediterranean. Drawing upon findings of the European research project WASSERMed,³ we consider a scenario at the year 2050, where supply and demand of water in the different countries are evaluated. Variations in agricultural productivity induced by the changing water availability are subsequently inserted into a global CGE model, to simulate the structural adjustment process for the world economy, and its implications in terms of virtual water trade. This section illustrates how the water availability scenario has been constructed.

The starting point is the current water balance for the 14 regional economies in our data set, that is the relationship between water supply (sources) and water demand (uses) at one period in time (e.g., a year). Among the supply sources, we distinguish between blue water and green water, supposing that only a fraction of total blue water is technically and economically accessible/exploitable. We estimate total water availability as the sum of accessible blue and green water, elaborating on data provided by Gerten et al. (2011).

Three uses of water are appraised: agricultural, municipal, and industrial. Green water can only be used in agriculture, where it is supplemented by blue water through irrigation or other means. Any difference between total blue water availability and consumption in the three categories above (where agricultural consumption is considered only for the part exceeding the green water stock) is interpreted either as unused water or water deliberately left for the preservation of aquatic ecosystems, which we refer to as environmental flow requirement (EFR). Water consumption by agriculture in the baseline (2000–2005) has been estimated using data from Chapagain and Hoekstra (2004). Municipal and industrial consumption has been obtained from the FAO—AQUASTAT database. Figure 5 shows the composition of water demand among different usage classes for the Mediterranean countries considered in this study.

Fig. 5

Composition of (blue) water consumption in the baseline (2000–2005)

To assess how much water will be available for agriculture in 2050, values for all supply and demand components have been projected to the future. On the supply side, total water availability has been estimated starting from data about future climate conditions, produced in WASSERMed from a set of regional and global circulation models. The climate scenario suggests that precipitation will generally decrease in the Mediterranean in the period 2000–2050, particularly in France (−13 %), Morocco (−18 %), and Tunisia (−10 %). The average temperature is expected to increase of about 2 °C. Climate scenario affects the future total water availability by means of changes in precipitation and temperature. Elaborating on these data,⁴ we predict a significant drop in water supply for France (−34 %), Italy (−14 %), Turkey (−11 %), Croatia (−10 %), Spain (−8 %), whereas countries in southern Mediterranean would be much less affected, because their water resources are relatively more independent from local climate conditions.

On the demand side, the various components have been projected using different assumptions and methodologies. For water consumption in agriculture, our reference point is a hypothetical situation in which agricultural production volumes stay unchanged. Of course, this is not meant to be a realistic scenario, but only a reference benchmark. Even with constant production levels, however, water demand would increase, because of higher temperature and evapotranspiration, of about 10 %.

Municipal consumption, that is water for human drinking, washing, etc., is generally assumed to follow demographic changes. Population projections have been taken from the World Population Prospect (United Nations Secretariat 2010), which devises a very strong growth of population in the Middle East. In addition, for some developing countries in our set, we consider the possibility that municipal water demand could increase more than proportionally, to reflect improved access to sanitation and freshwater.

Industrial consumption is assumed to increase at a rate equal to 1/3 of the national income growth. GDP forecasts have been derived from World Bank Statistics.⁵ The lower rate for industrial water consumption is intended to account for the changing composition of the national income, with a lower share for manufacturing industries, as well as improvements in efficiency.

In addition to water consumption, water resources may be needed to preserve a number of natural environments. The environmental flow requirement expresses this “pseudo-demand” for water as a share of total runoff, so we logically extend the notion to our estimates of blue water availability. The EFR concept itself is a rather elusive one, as there is no fixed threshold value for environmental preservation and much depends on collective evaluation. We look at the literature (Korsgaard 2006; Hirji and Davis 2009) to select some “reasonable values” for the EFR, which in this context means the share of blue water resources that should be set aside for effective protection of the environment.

We regard the EFR not as a constraint but as a policy variable. In other words, national governments may or may not be willing to save water for environmental purposes. In our numerical experiments, we assume that all countries in the European Union⁶ must comply with strict environmental regulation, so that the EFR share of (blue) water cannot be made available for consumption. Non-EU countries, on the other hand, are assumed to have more degrees of freedom, so they may opt not to comply with EFR requirements. In our scenario, we assume that non-EU countries do not comply with EFR requirements.

When municipal and industrial consumption and possibly EFR are subtracted from total blue water, what is left is water potentially available for the agricultural sector, supplementing green water. This “water potential” can be compared with estimates of agricultural water demand at fixed production levels. If potential water exceeds water demand, agriculture is not water constrained, at least if current production volumes do not significantly increase. Otherwise, (blue) water delivered to agriculture must be cut by a certain amount. Table 1 presents our estimates of reductions in water available for agriculture in 2050.

Table 1 Reductions in water available for agriculture in 2050 (only affected regions)

Five economies are found to have insufficient water resources, at the year 2050, to sustain current production levels in agriculture. Because of lower precipitation and higher temperature, France and Italy will be affected by a drop in water resources, with a larger impact for agriculture in Italy, because relatively more blue water is used there. Tunisia, Egypt, and Rest of Middle East/North Africa will also lack water for agriculture, but for different reasons. In Egypt and the Middle East, the climate change impact on total water availability will be negligible, as in this area much of the water is imported, pumped from the ground, desalinated, or recycled. However, non-agricultural water uses are expected to grow at a significant rate. Tunisia is a special case, since water resources are overexploited already in the baseline (see Fig. 5). Any further increase in water demand would not be sustainable.

We analyse how reductions in water availability could affect agricultural productivity. To this end, it is important to take into account that (i) each country has its own mix of agricultural products, and (ii) crops may differ in terms of sensitivity to water shortages. As in the GTAP database, we consider seven classes of agricultural products: wheat, cereals, rice, vegetables and fruits, oilseeds, sugar, and other products. For each crop group in each country, a “water elasticity” parameter has been estimated, accounting for both the physical characteristics of the crop and the overall efficiency of the water delivering system. Using these parameters, changes in water availability for agriculture have been translated into changes in agricultural productivity by sector, as reported in Table 2.

Table 2 Reduction in agricultural productivity by sectors and by region

On average, the impact on agricultural productivity is very high for Egypt, high for France and Italy, and medium for Tunisia and the Rest of Middle East/North Africa.

4 Future Virtual Water Trade in the Mediterranean

Estimates of Table 2 have been used to shock exogenous productivity parameters in a computable general equilibrium model of the world economy (Hertel and Tsigas 1997). A CGE model is a very large nonlinear system, which provides a systemic and disaggregated representation of national, regional, and multi-regional economies. The system includes market clearing conditions and accounting identities, to account for the circular flow of income and inter-sectoral linkages inside the whole economic system.

A simulation exercise entails comparing two equilibria for the global economy, in which all markets clear, before and after the variation of some exogenous parameters (in our case, multi-factor productivity in a set of agricultural industries). The model output includes all the main macroeconomic variables, such as nominal and real GDP, consumption and production levels, relative prices for products and primary factors. To summarize the overall macroeconomic impact of the simulated variation in agricultural productivity, Table 3 reports the estimated percentage variations for the national real income (which is a measure of aggregate welfare) for all regions in our set.

Table 3 Estimated variations in real national income

Not surprisingly, lower productivity in agriculture, induced by reduced water availability, generates negative consequences in terms of national income for most Mediterranean countries. The magnitude of the loss depends on the amount of the productivity shock, but also on the share of agricultural activities in the economy. Egypt is the country which is hurt the most, as the model estimates a fall of 7.24 % for real income. Significant reductions in GDP and welfare are also estimated for France, Italy, and Tunisia, which are the other water-constrained countries in our exercise. Three countries get (slight) benefits: Morocco, Spain, and Turkey. This is not because of improvements in productivity (which is unchanged there) but because of enhanced relative competitiveness vis-à-vis trading partners and competitors (a second-order general equilibrium effect).

The output of the CGE computer simulation comprises counterfactual estimates of trade flows. Using the same procedure applied for actual trade flows, illustrated in Sect. 2, it is possible to estimate virtual water flows for the scenario under consideration. Figure 6 is analogous of Fig. 2 and displays the most significant flows of virtual water between Mediterranean countries.

Fig. 6

Largest flows of virtual water trade in the Mediterranean in the counterfactual scenario (Mm³)

The most notable difference between Figs. 2 and 6 is that the amount of virtual water flowing from France toward Italy, Spain and, to a lesser extent, Morocco decreases significantly. On the other hand, imports of virtual water by France increase somewhat. The reason is easily found by looking at Fig. 7 (corresponding to Fig. 1 in Sect. 2), presenting estimates of the virtual water trade balance.

Fig. 7

Per capita net virtual water exports (balance) in the counterfactual scenario (Mm³)

In the current baseline (Fig. 1), the French virtual water trade balance was slightly positive. In the counterfactual scenario (Fig. 7), the balance gets negative, because the agricultural sector in France becomes much less competitive. Italy and Tunisia also deteriorate their virtual water trade balance, with less water flowing towards Spain and Italy, respectively. On the other hand, Morocco and Turkey grow in terms of virtual water exports.

Figure 8 shows from where the additional water imports come from, or to where additional water exports are directed. As such, this figure does not completely correspond to Fig. 2. We can see that France and Italy get most of the extra water imports from central and northern Europe. On the other hand, Spain increases its virtual trade exports towards the latter two countries.

Fig. 8

Variation in per capita virtual water flows (Mm³) for each Mediterranean economy

5 Concluding Remarks

This paper has provided some estimates of virtual water trade patterns in the Mediterranean. Virtual water trade follows conventional trade in goods (in this case agricultural products), whereas no water is physically exchanged. As international trade is a powerful mechanism for improving the allocation of economic resources, including water, the virtual water paradigm is just one way of looking at the potential benefits of international trade from a “water perspective”.

The Mediterranean is an area where water is scarce and unevenly distributed. Potential water demand exceeds supply in many Mediterranean countries, and problems are likely to be exacerbated in the future, because of climate change and reduced precipitation.

In this work, two cases have been considered: the current virtual water trade structure, related to trade in agricultural goods, and a future scenario, simulated by means of a computable general equilibrium model, where reduced agricultural productivity, induced by lower water availability, is taken into account.

Analysis of current virtual water flows reveals that most countries are net importers of virtual water, thereby realizing sizeable “savings”, particularly of precious blue water resources. Much of the intra-Mediterranean virtual water trade occurs between the largest northern economies (Spain, France, Italy) but, in per capita terms, the country which gets the largest amount of virtual water from abroad is Cyprus.

This picture will likely change in the time ahead, because the evolution of the world economy, as well as of international trade, will ultimately be reflected in varying virtual water flows. A simulation exercise has been performed in this paper, where we abstract from the many possible factors affecting future trade patterns, focusing instead only on the possible consequences of reduced water availability in the Mediterranean.

We found that both northern and southern countries will be affected by water shortages, although for different reasons. In the north, increased temperature and reduced precipitation will lessen water stocks. In the south, the driving factors will be demographic and economic development. Implications of this scenario in terms of virtual water entail reduction in intra-Mediterranean trade and increases in virtual imports from central and northern Europe, as well as from the rest of the world.

Appendix

See Table 4.

Table 4 Baseline virtual water trade flows (Mm³)