Analyzing Egypt water footprint based on trade balance and expenditure inequality

conducted a consumption-based analysis of water use in Egypt, using an interregional input-output (IRIO) model between Egypt and the Rest of the World (RoW). Achieving the first goal, Egy exports for 57 sectors of production are calculated. Our results indicate that Egypt saves 8% from its NWF through importing water-intensive products, especially wheat. However, Egypt virtually net-exported 12% of its total available freshwater lupine, medical and aromatic plants, henna) as well as paddy rice, and vegetable and fruits.


Introduction
In line with the UN 17 Sustainable Development Global Goals, in 2016, the Egyptian government launched the Sustainable Development Strategy 2030. The strategy contains a set of economic, social, and environmental goals to be accomplished by the year 2030. However, accomplishing these development goals may be restricted in the coming years by the challenge of water scarcity that Egypt suffers from.
Egypt is a very arid country: with only 1.18% of annual water resources comes from rainfall and 9.03% from underground sources, leaving the Nile river as the main reliable source of water in Egypt (CAPMAS, 2017). Under the 1959 Nile Waters Agreement between Egypt and Sudan, Egypt receives a constant amount of water (55,500 Billion cubic meters (m³)) through the Nile river annually (FAO, 2016). This amount of water accounts for 72.64% of the Egyptian annual water resources (CAPMAS, 2017).
On another hand, rapidly growing population; from 28.5 million in 1962 to 92 million in 2016; has increased the annual water demand to exceed its annual constant supply (FAO, 2017a). As a result, the total renewable water resources per capita in Egypt dropped from 2041 m 3 in 1962 to 637.1 m 3 in 2014 (CAPMAS, 2017;FAO, 2017a). Thus, according to Falkenmark Water Stress Indicator (Falkenmark, 1989), Egypt has reached a high and chronic water scarcity level that negatively affects human health, general wellbeing, and restricts the ability to accomplish economic development goals (Eckstein et al., 2010).
Moreover, based on demographic projections, the Egyptian per capita renewable water resources is projected to drop to 500 m 3 by the year 2030 (FAO, 2016). This limit accounts for the threshold of the absolute water scarcity in Falkenmark Indicator (World Bank 2016;Falkenmark 1989). This would seriously constrain the potential for accomplishing Egypt 2030 goals.
Although agriculture sector consumes 81.6% of Egypt annual water resources, (followed by family use (13.5%), and industry use (1.57%)) (CAPMAS, 2017), it contributes by 12% of Egypt`s annual GDP (World bank 2017). Significant changes in Egypt water-intensive crops production and trade have been observed during the last 50 years. For example, 570% increase in Egypt wheat production during the period 1961-2015 was observed and accompanied by 1456% increase in Egypt wheat net imports (FAO, 2017b). During the same period, 322% increase in Egypt rice production was observed and accompanied by 590% increase in Egypt rice net exports (FAO, 2017b).
Accordingly, due to the serious current state of Egypt water crisis, a more comprehensive perspective on the role of water in Egyptian trade and consumption is needed. Virtual Water (VW) trade on its freshwater water availability. This can be achieved through use on its water availability. This is done by calculating households NWF per product and comparing it with their direct water use. We also compare urban and rural NWF per product, according to their different patterns of expenditure. Thus, we have conducted the first top-ng an interregional input-output (IRIO) model between Egypt and the Rest of the World (RoW). The remainder of this paper is organized as follows. Section 2 clarifies literature review. Section 3 introduces the method. Section 4 provides data and their sources. Results are presented in Section 5. Section 6 introduces brief conclusions. A schematic overview of the study framework is clarified in (Figure 1).

Virtual water & water footprint
The virtual water concept was first introduced by (Allan, 1993). It is defined as the water embedded in agricultural products through their production and supply chain processes. Allan clarified that virtual water trade can be an effective tool for waterscarce regions to alleviate their water scarcity problem, through importing water-intensive crops rather than growing them domestically.
Nine years later, (Hoekstra and Hung, 2002) broadened the VW concept by introducing the concept of water footprint (WF). They defined the WF of a nation as the total amount of water used by the nation`s population. It is the sum of domestic water and the net virtual water imported from abroad, used directly and indirectly, to produce goods that satisfy population`s final demand, during a certain period. Thus, the concept of WF is much broader than the VW concept. It can be calculated by applying either a bottom-up approach (life-cycle analysis) or a top-down approach (Input-Output Analysis). The bottom-up approach was first applied by (Chapagain and Hoekstra, 2004). Using this approach, the WF of a product can be calculated by summing up the amount of water used directly 1 and indirectly 2 at each stage of production. However, this approach cannot 1 Water used in the production process directly. 2 Water used in the supply chain of inputs (during previous production stages). The IRIO analysis of WF and VW has been used by many researchers. (Carter and Ireri, 1970) developed an IRIO model to calculate the VW transfers between California and Arizona as well as the production per unit of water use in each region. They found that California is much more efficient in water use than Arizona. Recently, (Zhang et al., 2011) evaluated the total WF of Beijing and found that 51% of it is imported virtually, with the agriculture sector as the highest water consumer. (Suttinon et al., 2013) evaluated the water demand of each industrial sector in three regions of Japan, taking into consideration the impacts of implementing three different governmental water policies (reduce reuse recycle). (Deng et al., 2016) calculated China water footprint for the years 2002 and 2007 and compared them to analyze the changing trend of water use in China. (Chen et al., 2017) applied an IRIO model to estimate the WF of different provinces in China as well as the transfers of virtual water between those provinces. They found that the larger GDP and population, the larger water footprint for the province. They also found that many of the underdeveloped regions are suffering from water shortages. However, they export virtual water to the developed regions. (Zhang et al., 2017) studied the case of the scarcest water region in China, the North China Plain. They found that the water scarcity of this region is alleviated through virtual and actual interregional water transfers. According to the study results, the region is a net water importer through agricultural trade.

Egypt water footprint literature
During the last two decades, the Egyptian WF has been sparsely researched. (Wichelns, 2001) studied the comparative advantage of virtual water and the role of trading in it as well as trading in other limited resources to achieve food security and other national goals in Egypt. ) used a bottom-up methodology to calculate the virtual water content of agricultural products. They calculated national and global water losses and savings because of the international trade in these agricultural products. They also calculated Egypt blue water savings 3 resulting from imported wheat. (El-sadek, 2010) analyzed the VW concept as a solution for alleviating water scarcity in Egypt. He gathered different previous estimates for Egypt virtual water net imports according to food trade. In line with these results, (Zeitoun et al., 2010) analyzed Egypt virtual water net imports through trading in 17 crops and 4 kinds of livestock, which accounted for 54% of Egypt annual Nile River water flow during the period (1998)(1999)(2000)(2001)(2002)(2003)(2004). (El-gafy, 2014) investigated wheat production, water footprint, and virtual water nexus using a System Dynamic model. She found that the water footprint of wheat production and consumption in Egypt changes according to changes in the crop production, foreign trade, per capita consumption, population, and climate effects. In line with this study, (El-gafy et al. 2017) also used system dynamics to calculate a water food energy nexus index and the energy and water footprints for 43 Egyptian agricultural crops, based on production and consumption amounts. Also, they calculated the virtual water and energy imports and exports of the same crops.
Most of these previous studies have focused on WF of agricultural products. The inter-industry effects between all economic sectors may thus be underestimated in their analyses. Also, the effect of final consumption of products on water use has not been considered. Another gap in the literature is the understanding of effects of final expenditure differences related to different expenditure patterns on water use. This study aims to fill these gaps in the literature, employing an IRIO analysis of WF, disaggregated by expenditure profiles of rural and urban dwellers.

The IO framework
The traditional IO table describes the interindustry flows of products between economic sectors (from sector i as a seller to sector j as a buyer) in monetary units. Thus, products from sector i are distributed to satisfy the other sectors (interindustry) demand as well as the final consumer demand. This can be shown as follows: (1) ( 2) where is the total output of sector , is the number of sectors in the economy, is the interindustry flows (sales) from sector to all sectors (including itself, when ), and is the final demand for sector products. Using the matrix form, let: ( 3) where is the total output vector of sector i, is the intermediate demand matrix on products (the inter-industry transactions), and is the final demand vector on products. We can obtain the technical coefficient matrix (denoted also as direct requirement matrix) by dividing each by the total output in each corresponding row. Each element in the matrix represents the monetary unit worth of inputs from sector needed to produce one monetary unit worth of output of sector . Thus, each element in can be replaced by . Accordingly, (1) can be represented as follows: (4) In matrix notation, ( ) can be represented as follows: where: is the identity matrix and is the Leontief inverse matrix (denoted also as the total requirement matrix), which is represented as follows: (6) where: denotes the monetary unit worth from sector products that are necessary to satisfy one monetary unit worth of final demand of sector products. Thus, the link between final demand and corresponding direct and indirect production is built with this Leontief inverse matrix.
, are the Egyptian final demand for sector i goods that are produced in Egypt and RoW respectively.
, are the Row final demand for sector i goods that are produced in Egypt and RoW respectively. The technical coefficients matrix for IRIO model is expressed in the following partitioned matrix: Thus, the formula in equation (5) can be represented in the following partitioned matrices (10) where: is the Leontief invers matrix in the IRIO model.

Water consumption multipliers
Consider a row vector of the annual domestic freshwater consumed by production sectors in Egypt and RoW as follows: where , and are the annual domestic freshwater consumed directly by j production sectors in Egypt and RoW 8 respectively. The water extended IRIO model can be shown in (Figure 2)   The figure depicts the interindustry sub-matrices: , Egypt final demand sub-columns, RoW final demand sub-columns, total output sub-columns , the value-added row; that presents the other payments from production sectors to the non-industrial inputs of production, such as wages, depreciation of capital, or taxes; and finally the freshwater row . Now, it is possible to calculate water intensity vector (also called water direct impact coefficient vector) as follows: (12) Each element in and specifies the amount of domestic/RoW water required directly per one monetary unit of total output of sector production in Egypt/ROW respectively. By rearranging (12) we have (13) Recalling back equations (5) and (6), we can simplify (13) to: This formula was first introduced by (Just, 1973). We use this formula to clarify the total impacts of final demand for water use. We can obtain the indirect water use per one monetary unit of total output (water embodied in the intermediate inputs of production) by subtracting from . Also, we use the diagonalized to show the source sector and region of water embodied in all products as follows: Let defines the water total impact coefficients matrix. It is also known as the water intensity multiplier matrix. Each element of this matrix reflects the amount of water required directly and indirectly to generate one monetary unit of final demand on sector j. We can express in the IRIO model in the following partitioned matrix: where: , and express the intraregional effects for changes in domestic final demands, and and express the interregional spillover effects in one region that are caused by changes in another region`s final demand. Then, we can obtain the water consumption multipliers as follows: (17) where: , and are the intraregional water consumption multiplier vectors for Egypt and RoW respectively. Each element in each vector represents the total amount of water used in all sectors in a specific region (Egypt or RoW) to satisfy one monetary unit of final demand for sector output of the same region (Egypt or RoW).
, and are the Interregional water consumption multiplier vectors. Each element in each vector represents the total amount of water used in all sectors in a specific region (Egypt or RoW) to satisfy one monetary unit of final demand for sector in the other region (RoW or Egypt). vector.

National Water Footprint model and traded water
According to , the National Water Footprint of a country consists of internal and external WF. The internal WF is the amount of domestic water used to produce products that are consumed domestically. It can be calculated as follows: (18) where: is the domestic water embodied in Egypt exports to the RoW industries, and then, re-imported to Egypt through Egyptian imports to On the other hand, the external WF is the VW imported into the country to satisfy annual final demand. It can be calculated as follows: where is the amount of virtual water embodied in products that are transmitted from to Egypt, is the RoW water embodied in imports to Egypt intermediate demand, is the Row water embodied in imported final products Accordingly, the NWF by sector is presented as follows (20) According to this formula, each element in the vector attributes the total amount of water demand to the source sector and region where the water was directly consumed. This formula distinguishes between domestic and foreign water. Exports of virtual water are calculated as follows: On the other hand, we can calculate the Egyptian NWF by product as follows: According to this formula, each element in the vector attributes water consumption to the final product it becomes embodied in. This formula does not distinguish between domestic and foreign water use, but identifies the virtual water embodied in final demand of each product .  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 3.5 National Water Footprint for different expenditure patterns Introduced by (Miyazawa, 1976), final expenditure is separated into distinct groups according to their different income classes. Due to limited expenditure data across income groups for Egypt, we attempt to make a more simplified analysis than Miyazawa`s. Using two Egyptian final expenditure patterns; for rural and urban geographical areas; we can calculate National water footprint for each area as follows: where: and represent national water footprint for urban and rural areas respectively. and represent Egypt final demand on sectors products for urban and rural households respectively.

Economic data
The method and data sources have been adapted from Holland et al. (2015), who integrated the hydrological model Water Gap with a multi-region IO model. The transactions matrix was extracted from The Global Trade Assessment Programme (GTAP) for the year 2007 (Peters et al., 2011) as it has a disaggregated agricultural sector. The original GTAP table includes 113 regions with 57 sectors in each region. We have merged the 113 regions into only two regions: Egypt and the RoW. However, we have kept the 57 GTAP sectors without merging. Accordingly, the technical coefficients matrix (A) in this study has a dimension of 114 rows by 114 columns. Also, the RoW final demand data was obtained from the same data source (Peters et al., 2011).
With respect to the final demand in Egypt, we used two classifications of data for two purposes of calculations. First: we used the total Egyptian final demand (contains households and government) to calculate the WF of final demand of each product. Data of total Egyptian final demand are obtained directly from the (GTAP) database (Peters et al., 2011). Second: we have distinguished between household final demand in rural and urban areas to calculate the WF by product of each area. The final demand data for each of these areas are not available directly, and hence need to be derived. Firstly, we directly obtained the 2008 average final expenditure per family for urban and rural areas from the Egyptian Statistical yearbook of the Egyptian Central Agency for Public Mobilization and Statistics (CAPMAS, 2015) 9 . Secondly, we calculated the total final expenditure for each classification as follows:  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 The calculated final expenditure different product groups in Egypt. We have allocated these product groups to the 57 GTAP sectors according to the GTAP final demand ratios per sectors 10 . Finally, we proportionated these allocated final expenditures to the GTAP households final demand data to obtain the final demand for urban and rural areas in Egypt as follows:

Hydrological data:
The study accounts for only the freshwater consumption which is defined as the amount of freshwater that is completely withdrawn from the system and not returned. We obtained the RoW freshwater consumption data from (FLÖRKE et al., 2013;Holland et al., 2015).
With respect to Egypt freshwater consumption, we distinguished between two categories of data: (1) the agriculture water use data, and (2) the other sectors water use data. We obtained the detailed 2007 agriculture water use data per 30 crops in Egypt directly from the CAPMAS Irrigation and water resources statistics report (CAPMAS, 2008). Then, we aggregated these data with the 8 GTAP crop categories. See supplementary material, (Table S2).
With respect to the other sector`s water use data, it is defined in this study as the amount of water consumed by all sectors except that which is consumed by agriculture and households. Thus, we obtained these data from the CAPMAS annual report (CAPMAS, 2011) by subtracting the amount of water consumed by agriculture and households from the total amount of water consumed by all sectors.
Because of the non-availability of data about how each sector consumes water from the other sectors, we have split the other sectors water use of Egypt across the GTAP sectors except agriculture ones, based on what each of these GTAP sectors spends on water, as done in (Holland et al., 2015). The more a sector spends on water, the higher the proportion of the water use it is assigned. We get the Egyptian sectoral spending on water from the 2007 GTAP input-output transaction matrix (Peters et al., 2011).

Results and discussions
5.1 Egypt NWF and VW trade balance 3 . 11 According to (figure 3), the NWF of Egypt is exceeded by its direct water consumption (W e ), which means that Egypt is a net VW exporter. Thus, Egypt VW net exports is calculated to be 8.4 Billion m 3 in 2007 which accounts for 12% of Egypt total water resources in the same year (70 Billion m 3 (CAPMAS, 2011)). Egypt VW net export for each 57 production sectors are presented in (Figure 4) 12 and aromatic plants, henna) has the largest amounts fruits. Wheat has the largest amounts of Egypt VW net imports. Thus, water intensities of products should be more considered while putting the Egyptian international trade policy, especially exporting water-intensive crops such as feed plant, alfalfa, and rice. 13   Egypt households NWF excluding government NWF) accounts for 7.8 folds of the family direct water use (they consume 89% of water virtually and 11% directly). The latter is compared with the top 25 NWF products in (Figure 5). It shows that the water consumed directly accounts for 14% of the water embodied in edible products that they consume annually. Changing diets and monitoring their consumption of such water-intensive food crops can contribute to conserving Egypt water resources.

Urban/ rural households water use
Using two Egyptian final expenditure categories for urban and rural areas, NWF product is calculated for each area (government`s NWF is excluded). Results of 57 products are summed up into 23 product groups and presented in (Table 1). Because of the larger population, total NWF product for the rural exceeds that for the urban. However, the average NWF product per capita for the urban is 16.6% higher than that for the rural. This is due to the following reasons: (1) different consumption patterns between urban and rural. The cattle and animal product, for example, accounts for 18% of the urban NWF product percapita, while it accounts for 13% of the rural NWF product per-capita. 14 (2) different intraregional (domestic) water consumption multipliers for Egypt products; where edible products have the highest fractions. 15 Thus, we can conclude that reducing urban and rural intakes from domestically water reliant products that they consume heavily (agricultural, cattle and animal products for urban, and agricultural and other food for rural) can significantly contribute to reduce their NWF and conserve the country s domestic water.
14 Fractions of NWFproduct per-capita of 23 product group for urban & rural populations are presented in (Figure S2) in the supplementary material. 15 For further detailed results about the intraregional and interregional water consumption multipliers for 57 Egypt products, see supplementary material, (Table S3)

Conclusion and implications
The conservation or depletion of a country`s water resources can be significantly affected through its international trade expenditure on goods and services. Due to the serious current state of Egypt water crisis, this study had two fundamental goals: [1] Tracing the effects of Egypt Virtual Water (VW) trade on its real water availability; and [2] analyzing the effects of hou conducted a consumption-based analysis of water use in Egypt, using an interregional input-output (IRIO) model between Egypt and the Rest of the World (RoW). Achieving the first goal, Egy exports for 57 sectors of production are calculated. Our results indicate that Egypt saves 8% from its NWF through importing water-intensive products, especially wheat. However, Egypt virtually net-exported 12% of its total available freshwater lupine, medical and aromatic plants, henna) as well as paddy rice, and vegetable and fruits.
product, according to patterns of expenditure. Our results indicate that (a) the water embodied in edible products that households consume annually is more than sevenfold their direct water consumption, and (b) the NWF per capita is roughly 17% higher for the urban population than for the rural. This is because urban citizens consume larger quantities of almost all products. The water embodied in household consumption in Egypt can be seen to be exacerbating the water scarcity crisis. This effect is particularly seen in the consumption of water-intensive agricultural and animal products.
Our results show the need to monitor the virtual flows of freshwater when exploring freshwater scarcity. Focusing on direct water use, as is conventional practice, ignores the role of trade as a driver of water shortages. Countries must consider both the territorial and international demand for freshwater resources to enhance both our understanding of the security of food supply (which is very water intensive), and broader issues of sustainability through the link between freshwater resources, human well-being, and economic development. Embodied water calculations moreover open up different policy avenues to reduce risks of water shortages. Embodied water use can enable the identification of opportunities to reduce the indirect reliance on scarce water resources, such as how diets or fiscal incentives on water, food and other resources drive freshwater use. By understanding the impact of different socio-demographics (in our case urban and rural), we can explore the distribution of resources across the population and identify demandfood and water.
In Egypt, applying trade policies that stimulate water-intensive imports and reduce their exports, such as pricing irrigation water rather than totally subsidizing it, may have an effect. However, policies need to be evaluated with caution, as they might have important social impacts. Directing household choices towards lower water-intensity products may alleviate water scarcity. However, since household choices are affected by diverse factors, including their awareness, tastes, income, product prices, geographic location, and diet/culture, several complementary specific policies may need to address each of these aspects. Product labelling, for instance, may be a good tool for raising awareness, but alone will not be sufficient to address the issue. Complementary upstream policies, such as incentives subsidizing lower water-intensity products, rather than their thirsty counterparts, and reflecting water costs in the price of water-intensive products, may have a larger effect on changing consumption patterns but should be applied with caution to avoid hindering development.
ion. We chose to use the GTAP-Water GAP model developed in Holland et al. (2015) as it treats the agriculture sector as multiple sectors with different freshwater use requirements, however, the latest year available was 2007. Other IO databases were deemed unsuitable: for example, EORA does not have a disaggregated agricultural sector, and EXIOBASE is a European-centric model and therefore does not represent Egypt as a standalone country.
An up-to-date IO table would clearly improve this study -however, GTAP is not updated on an annual basis, making it difficult to capture dynamic changes. On the other hand, water use, and availability are constantly influenced by dynamic changes in both the economy and natural circumstances. For example, floating the exchange rate, an economic policy action taken at the end of 2016, affects the Egyptian trade balance and indirectly its water resources. Those effects could be traced in further analysis using an updated IO table for the time following the implementation of this policy. Further research could domestic water footprint. This could open avenues for policymakers in Egypt to apply more socially inclusive water policies .  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63 64 65     Egypt saves 8% of its National Water Footprint through imports. Egypt virtually net exports 12% of its total available freshwater resources. Households consume 11% of water directly and 89% virtually. Urban households national water footprint per capita is 17% higher than rural s. Water Footprinting can be used to make national policy recommendations.