Abstract

This paper aims to investigate the relationship between virtual water (VW) exports and crop exchange by employing the methodology of social network analysis (SNA). This descriptive analysis gives prudence for policy-makers about both central importers and influential exporters of VW using the degree and eigenvector centrality measures. In addition, to facilitate the communications between trading partners, each of them should reach the others with the fewest number of links, so, the small world network properties could be examined. This approach is applied on the yearly average VW exports of the Nile basin countries over the period 2000–2013, and some insights for VW exchange structure are investigated. The empirical results show that all Nile basin countries do not suffer from vulnerable VW export structure. They have a stable and balanced crop export structure. Kenya, Uganda, and Tanzania are identified as the most influential and effective countries in exporting VW of crops. The presence of these countries is unavoidable in drawing trade policy and water management plans. While Kenya succeeded in saving a significant amount from VW export network, Tanzania, Uganda, and Ethiopia are gaining losses. Furthermore, VW export network of crops among Nile basin countries satisfies the conditions of small world effect.

Introduction

Many studies on virtual water trade (VWT) have been done at different spatial scales, ranging from global scale to national scale. Hoekstra & Hung (2005), Chapagain et al. (2006), Hoekstra (2010), and Dalin et al. (2012) have conducted research at the global scale. Hoekstra & Hung (2005) have shown that about 13% of the water demand in the world is not directed for domestic crop production but for export purposes in the virtual form. Allan (1998), Shuval (2007), El-Sadek (2010), and Shi et al. (2014) have studied VWT at the national scale, and reached the conclusion that both water-poor and water-rich economies can benefit from VWT. Water-poor economies can overcome water scarcity by importing water-intensive crops instead of producing them domestically, while water-rich economies can benefit from VWT by exporting virtual water (VW) from their abundant water resources to maximize their economic returns.

Zeitoun et al. (2010) and Fattah et al. (2016) have analyzed the concept of VWT in Nile basin countries1. Zeitoun et al. (2010) have estimated the initial approximation of VWT for Nile basin countries from ‘water scarcity’ point of view. Fattah et al. (2016) have developed a mathematical optimization model to check the possibility of improving the agriculture water use and decreasing food gap. However, there is a lack of research in the network structure of VWT among Nile basin countries, how much water each country can save as a result of exchanging crops, and which of them are vulnerable or influential in VWT network. Such investigations could lead to sustainable trade practices and water management plans. Guided by the studies of Shi et al. (2014) and Lee et al. (2016), this paper aims to investigate this gap and gives some insights in this direction.

The paper analyzes the characteristics of VW export network of crops among Nile basin countries by applying the methodology of social network analysis (SNA) for the yearly average of VW exports over the period 2000–2013, and obtains some insights for VWT flow structure. This descriptive analysis gives prudence for policy-makers about both central importers and influential exporters of VW using degree and eigenvector centrality measures. For the importers, the study determines how vulnerable each importer of VW for the selected crops2 is, and how much the amount of water that each importer gains through intra-trade of these crops. Adger (2006) proposed a framework to review the analytical approaches of vulnerability in relation to risk analysis, disaster management, and climate change. Fraser et al. (2012) defined vulnerability as the sensitivity of a system (social or natural system) to be harmed and unable to absorb or adapt to any stresses and exposures. Furthermore, in the context of VWT, Lee et al. (2016) argued that the country which imports a significant amount of VW from a few number of exporters is a vulnerable country. Its trade structure will be vulnerable and non-resilient as a result of being highly dependent on the exporters to increase its local food supply.

Sartori & Schiavo (2015) contributed to the definition of vulnerability by linking between the structural features of the global network of VWT and the resilience of the global system to food supply shocks. They showed that the effect of food supply shocks has increasing negative impact on the importing country if the network of VWT is dominated by a few exporters. In this paper, we follow the definition of Lee et al. (2016) and Sartori & Schiavo (2015) in the context of structural features of VWT network to determine which Nile basin country has a vulnerable or non-vulnerable trade structure.

In addition, the study determines which Nile basin countries are more influential in VW exports, and hence being good export partners for the selected crops in the future. To facilitate the communications between these countries, each of them should reach the others with the fewest number of links, so, the small world network properties are examined for this purpose.

To sum up, three main objectives are specified for this study. The first is to evaluate the vulnerability of each importing Nile basin country using in-degree and weighted in-degree centrality measures. The second is to analyze the influential export partners of VW export network of the selected crops using the eigenvector centrality measure. The third is to examine the presence of small world effect in the VW export network of Nile basin countries.

This paper is organized as follows: the next section below gives the criteria for choosing the basket of crops, a brief introduction about VWT and how to calculate its values, the formulation of the VW export network of Nile basin countries, and the centrality measures which are used to construct our analysis. This is followed by a section describing the results in the form of the properties of VW export network of Nile basin countries, the vulnerable importers, national water savings, the most influential exporters, and the effect of small world phenomenon. The final section is devoted to the main conclusions and policy implications of the results.

Methods

Selection of crops studied

In order to construct the crop profile of VW export network of Nile basin countries, we selected bananas, rice, beans-dry, maize, potatoes, and wheat crops. These crops are the major food crops produced in Nile basin countries. They, in total, represent about 15% of the total crop and livestock exports for Nile basin countries as indicated by FAOSTAT3. Maize, rice, and wheat are among the major cereals in the Nile basin and are used for approximation for trade balance for these countries, as reported by the official website of Nile information system4. Furthermore, these crops consume about 27% of green water for their production in Nile basin countries, as Zeitoun et al. (2010) indicated.

Virtual water content

The term virtual water content (VWC) is defined, according to Hoekstra & Chapagain (2008), as ‘the volume of water used to produce a product’. VWC is divided into two components: blue and green water. According to the water footprint network5, the former includes surface and groundwater, in other words, the water in freshwater lakes, rivers, and aquifers, and the latter is the precipitation on land that does not run off or recharge the groundwater but is stored in the soil or temporarily stays on top of the soil or vegetation. Eventually, this part of precipitation evaporates or transpires through plants. Many studies have based their analysis on both green and blue water contents of agricultural products. However, the present analysis is focused on green VWC to construct the network analysis for the following reasons. Mohamed (2016) claims that except for Egypt and Northern Sudan, the rest of the Nile basin countries do not depend on blue water in their agriculture. He stated that over 87% of cultivated land in the Nile Basin is under green water agriculture. Zeitoun et al. (2010) also reported that the green water component for a crop is close to zero in both Egypt and Northern Sudan. Accordingly, these two countries will appear in our study only as importers of green VW. In addition, we assume that the green VWC is constant over time and its estimates for the selected crops are as given in Table 1.

Table 1.

Virtual water content of the selected crops – green water content (m3/ton).

Crops Burundi Congo Egypt Eritrea Ethiopia Kenya Rwanda Sudan Uganda Tanzania 
Bananas 1,960 2,854   592 2,397   2,539 1,169 
Rice  6,120     3,176  4,178 4,181 
Beans-dry 2,954 4,077  4,180 2,999 6,523 4,470 3,603 5,585 5,432 
Maize 2,946 3,691  5,321 1,755 2,318 3,197 8,406 2,727 2,664 
Potatoes 1,112 509   484 752 567  498 293 
Wheat 3,462 1,339  2,118 2,021 1,225 3,045  1,319 1,907 
Crops Burundi Congo Egypt Eritrea Ethiopia Kenya Rwanda Sudan Uganda Tanzania 
Bananas 1,960 2,854   592 2,397   2,539 1,169 
Rice  6,120     3,176  4,178 4,181 
Beans-dry 2,954 4,077  4,180 2,999 6,523 4,470 3,603 5,585 5,432 
Maize 2,946 3,691  5,321 1,755 2,318 3,197 8,406 2,727 2,664 
Potatoes 1,112 509   484 752 567  498 293 
Wheat 3,462 1,339  2,118 2,021 1,225 3,045  1,319 1,907 

Source: Extracted from Zeitoun et al. (2010), Table 2.

Virtual water trade and its network

VWT is related to the trade of agricultural commodities, and consists of VW imports and exports, as Shi et al. (2014) demonstrated. To quantify the term VWT, Hoekstra & Chapagain (2008) suggested the following formula: 
formula
(1)
where VWT is the VW volume exported of the product p from the exporting country to the importing country in year t. PT is the trade volume from the exporting country to the importing country. VWC is the total water quantity needed either to produce the product p in the exporting country or to consume it in the importing country. Trade data for the period investigated are obtained from FAOSTAT's detailed trade matrix. As a first step of computation, the annual VWT volumes are calculated for each country by using Equation (1), where for each crop p, its VWC, given in Table 1, is multiplied by its total annual amount of exports PT. Then, after summing for all traded crops, the average of the annual VWT volumes is calculated for the whole period of study. This process is repeated for each trading partner in the Nile basin. The third column in Table 2 gives the resulting estimates of the yearly average of green VW exports.
Table 2.

The estimated annual average of green VW exports of the selected crops in Nile basin countries.

Exporting country Importing country Average VW export (Hm3/ton) 
Burundi Congo 0.002 
Kenya 0.100 
Rwanda 0.291 
Tanzania 0.062 
Ethiopia Egypt 3.484 
Kenya 35.137 
Sudan 5.294 
Tanzania 9.772 
Kenya Burundi 0.039 
Congo 0.022 
Egypt 0.174 
Eritrea 0.147 
Ethiopia 0.523 
Rwanda 0.300 
Sudan 1.199 
Uganda 0.896 
Tanzania 2.305 
Rwanda Burundi 3.390 
Congo 3.069 
Kenya 2.401 
Uganda 2.108 
Tanzania 1.087 
Uganda Burundi 13.262 
Congo 20.852 
Egypt 0.005 
Kenya 27.756 
Rwanda 8.835 
Sudan 10.512 
Tanzania 19.388 
Tanzania Burundi 6.681 
Congo 3.734 
Egypt 0.210 
Kenya 15.926 
Rwanda 9.565 
Sudan 0.059 
Uganda 36.507 
Exporting country Importing country Average VW export (Hm3/ton) 
Burundi Congo 0.002 
Kenya 0.100 
Rwanda 0.291 
Tanzania 0.062 
Ethiopia Egypt 3.484 
Kenya 35.137 
Sudan 5.294 
Tanzania 9.772 
Kenya Burundi 0.039 
Congo 0.022 
Egypt 0.174 
Eritrea 0.147 
Ethiopia 0.523 
Rwanda 0.300 
Sudan 1.199 
Uganda 0.896 
Tanzania 2.305 
Rwanda Burundi 3.390 
Congo 3.069 
Kenya 2.401 
Uganda 2.108 
Tanzania 1.087 
Uganda Burundi 13.262 
Congo 20.852 
Egypt 0.005 
Kenya 27.756 
Rwanda 8.835 
Sudan 10.512 
Tanzania 19.388 
Tanzania Burundi 6.681 
Congo 3.734 
Egypt 0.210 
Kenya 15.926 
Rwanda 9.565 
Sudan 0.059 
Uganda 36.507 

Source: Calculated.

The relationship of VW intra-trade among Nile basin countries can be represented as directed weighted network. Each Nile basin country represents a node in the VW export network, VW export flows corresponding to the selected crops represent the links between nodes with a direction pointing from one node representing VW exporter, and the direction pointing to a specific node representing VW importer. The volume of VW exports between two nodes is the weight of the link.

Degree and eigenvector centralities of the VWT social network

SNA is a distinguishable technique as Wasserman & Faust (1994) presented; it focuses on the structural or relational information to study and evaluate theories. SNA is concerned with data of ties among the actors. This does not mean that it ignores the actors in analysis, but its first and main concern is the relationships among actors. In this sub-section, the centrality measures used to evaluate the central importers and exporters are briefly reviewed. Freeman (1977, 1978) and Kosorukoff (2011) defined degree centrality measure as the number of links each node has within the network. Wasserman & Faust (1994) distinguish between in- and out-degree centrality measures. In-degree attributes to the number of importers and out-degree attributes to the number of exporters in the VW export network context. More precisely, these two centrality measures are calculated for each node by Equations (2) and (3) as follows: 
formula
(2)
 
formula
(3)
where n is the total number of nodes in the network. From these equations, a high score of in-degree for any Nile basin country indicates that it imports from various exporters, while a high score of out-degree indicates that it exports to many importers. The above equations can be extended directly for weighted directed graph by Equations (4) and (5) as follows: 
formula
(4)
 
formula
(5)

In this case, high score of weighted in-degree for any Nile basin country indicates that it imports a greater volume of VW, while a high score of weighted out-degree for any Nile basin country indicates that it exports a high volume of VW to other Nile basin countries.

Some Nile basin countries may affect the entire network structure of VWT in the region significantly. Thus, it is important to specify these countries. To do that, eigenvector centrality measure is calculated to find out the most influential exporting countries. Kosorukoff (2011) elaborated upon what Bonacich said in 1972 about eigenvector centrality measure and proposed Equation (6) to calculate this measure: 
formula
(6)
where A is the square adjacency matrix, and any entry of this matrix equals 1 if the relationship between node i and node j is present, and 0 otherwise. And, is the degree centrality scores vector of the node neighbors. Finally, λ is a constant which is known as the eigenvalue associated with the eigenvector

Small world network

Kleinberg (2004), Easley & Kleinberg (2010), and Newman (2010) discussed the experiment conducted by Milligram and his colleagues in the 1960s. The core of this experiment was that a set of individuals are connected through a few links. These individuals can reach each other only through six steps of links on average, which was called ‘six degrees of separation’. The experiment was conducted over 296 randomly chosen ‘starters’ to forward a letter to a ‘target’ person who lived in Boston. The address and occupation of the target person were given to the starters. The steps of this experiment, that the letter was passed through a chain of friends to eventually be close enough to the target person, were observed. In other words, the letter was passed by acquaintance to acquaintance within a social network, until it was reached in a few number of links or steps.

Two parameters are specified to consider the network a small network, as Estrada (2011) identified: clustering coefficient and average path length. The network will be considered ‘small’ if we get a high average clustering coefficient and a small average path length. The values of these two parameters are calculated by Equations (7) and (8) as follows: 
formula
(7)
 
formula
(8)
where and are the average clustering coefficient and the average path length, respectively, while k is the degree centrality measure, and n is the network size.

Results and discussion of green VW export network

Properties of green VW export network in Nile basin countries

In this sub-section, we employ both degree and weighted degree centrality measures given by Equations (2)–(5), for the yearly average green VW export network of the selected crops in Nile basin countries over the period 2000–2013. The resulting graphs which summarize the main results of the analysis are presented in Figure 1. In these graphs the node size indicates in-degree and weighted in-degree score such that a larger node is a more central importer. Label size reflects out-degree and weighted out-degree score such that a larger label is a more central exporter. A thicker arrow indicates a bigger green VW export flow between the two nodes it connects, while a thicker head of the arrow directed to a specific node indicates a higher volume of green VW export received by that node.

Fig. 1.

The yearly average virtual water network of crop exports over the period 2000–2013 in Nile basin countries: (a) in-degree, (b) out-degree, (c) weighted in-degree, and (d) weighted out-degree. Figures produced by Gephi 0.9.2.

Fig. 1.

The yearly average virtual water network of crop exports over the period 2000–2013 in Nile basin countries: (a) in-degree, (b) out-degree, (c) weighted in-degree, and (d) weighted out-degree. Figures produced by Gephi 0.9.2.

According to the out-degree centrality, as calculated by Equation (3), Kenya, Uganda, and Tanzania are the most central exporters of green VW export of the selected crops in Nile basin countries because they have the largest numbers of connections with various importers. Although Kenya is classified as the largest exporter in the region in terms of the number of connections, it only exports a small amount of green VW to these many importers. It exports 5.6 Hm3 to other countries. This implies that Kenya has many sources to obtain foreign currency. If any of its importing partners faces currency shock, Kenya can live with this crisis by exchanging with others. In addition, a small amount of exported VW to other Nile basin countries may allow Kenya to save water for national water usage and to satisfy the domestic needs for water.

Uganda, Tanzania, and Ethiopia export large amounts of green VW as the weighted out-degree centrality calculated by Equation (5) shows. These countries play the main role for water supply in the form of green VW exports in the Nile basin region, and this exporting role stresses their available water resources for exporting purposes. This result agrees with Zeitoun et al. (2010), who also found that Uganda and Tanzania are among the three biggest exporters of VW of crops in the Nile basin.

To evaluate the vulnerability of green VW importers, in-degree and weighted in-degree centrality measures, as given by Equations (2) and (4), are calculated. To classify Nile basin countries and identify which of them has a vulnerable or non-vulnerable crop trade structure, the scores of both in-degree and weighted in-degree measures are used to conduct this classification. Figure 2 shows the scores of both in-degree and weighted in-degree centrality measures for Nile basin countries.

Fig. 2.

In-degree centrality scores (a) and weighted in-degree centrality scores (b) for Nile basin countries for the yearly average of green VW export network over the period 2000–2013.

Fig. 2.

In-degree centrality scores (a) and weighted in-degree centrality scores (b) for Nile basin countries for the yearly average of green VW export network over the period 2000–2013.

From the in-degree centrality measure classification, Kenya, Egypt, Tanzania, and Congo are the largest importers in terms of the number of links. Meanwhile, Kenya is the only large importer for high amounts of green VW of crops in Nile basin based on its score of weighted in-degree centrality measure. It imports about 81. The same country is also among the two biggest importers of VW of crops in the Nile basin calculated out by Zeitoun et al. (2010).

The two-way Table 3 classifies the importing countries by in-degree centrality scores into low, medium, and high-connectivity groups, and by weighted in-degree centrality into small, medium, and large-volume groups. For in-degree centrality, importers who have less than one-third of the maximum in-degree score are classified as low-connectivity6 group. Importers who have scores between one-third and two-thirds of the maximum score are classified as the medium-connectivity group. And, importers who have scores greater than two-thirds of the maximum in-degree score are classified as the high-connectivity group. Similar classification rules are applied to classify the importing countries according to their calculated weighted in-degree centrality scores.

Table 3.

Classification of importers based on in-degree and weighted in-degree centrality scores of green VW export network of the selected crops in Nile basin countries.

 In-degree centrality scores 
Weighted in-degree centrality scores  Low (A) Medium (B) High (C) 
Small (I) ETH, ERI  BDI, SDN, RWA, EGY 
Medium (II)  UGA TZA, ZAR 
Large (III)   KEN 
 In-degree centrality scores 
Weighted in-degree centrality scores  Low (A) Medium (B) High (C) 
Small (I) ETH, ERI  BDI, SDN, RWA, EGY 
Medium (II)  UGA TZA, ZAR 
Large (III)   KEN 

Accordingly, countries, having a score of in-degree less than 1.7, belong to low-connectivity group, those having a score greater than 3.33 belong to high-connectivity group, and the rest who have scores of in-degree between 1.7 and 3.33, are classified in the medium-connectivity group. On the other side, countries who import less than 27 belong to small-volume group, while countries who import greater than 54 are classified in large-volume group. Medium-volume group includes the countries who import VW between 27 and 54 From Table 3, green VW export structure does not indicate vulnerability of any of the countries in the network according to the adopted definition of vulnerability as discussed in the introductory section. Ethiopia and Eritrea belong to the A-I group; they import small amounts of green VW from a small number of exporters. Burundi, Sudan, Rwanda, and Egypt belong to the C-I group. They import a small amount of green VW from a large number of exporters. Thus, they have diversified export partners. Uganda belongs to the B-II group which means that it imports a medium amount of green VW from a medium number of exporters. Tanzania and Congo are placed in the C-II group. They import a medium amount of green VW from a large number of exporters. Finally, Kenya stands alone in the group C-III as it imports a greater amount of green VW from a large number of exporters. Being the largest importer in terms of VW volume and the number of links allows Kenya to have various exporting countries and not to be dependent on a single country's political and economic conditions. In other words, if one of the exporting countries of VW to Kenya faces any food or water crisis or imposes any export bans on Kenya, Kenya has the ability to absorb this shock by importing from another country with which it is connected. Diversification of import sources for Kenya prevents the exporting countries playing a monopolistic role on Kenya's food supply. This finding frees Kenya from the debate mentioned by Marchand et al. (2016) in the context of negative effect of trade dependency on the exporting countries.

To sum up, all Nile basin countries have a stable and balanced green VW export structure. Each Nile basin country can absorb and cope with any potential food shocks and water crisis in any exporting country with which it is connected. If this situation continues, Nile basin countries may not face severe water shortage problems in the future.

The impact of VW exports on water saving of the Nile basin

Nile basin countries can benefit from VWT to save water resources through importing water-intensive agriculture products. If the country, instead of importing crops, cultivates them domestically, it will need additional water supply to do that. This will stress heavily on the internal water availability in that country. Thus, we are interested in calculating national water saving of the importing countries due to exchanging the selected crops under study by applying the following equation proposed by Hoekstra & Chapagain (2008): 
formula
(9)
where v is the VWC of product p in country, is the imported volume of product p, and is the exported volume of product p. A positive value resulting from Equation (9) is interpreted as an amount of water saved from VWT, while a negative value is interpreted as an amount of water lost from VWT. In this context, when the country exports more VW than it imports, this means that the country stresses on its available water supply. The term ‘loss’ means that the water used for producing agricultural products that are exported is not available anymore for domestic purposes, as mentioned by Chapagain et al. (2005).

Figure 3 compares between national water saving of green VW from crops' exchange and the average water required to cultivate these crops domestically over the period 2000–2013. Tanzania, Uganda, and Ethiopia are gaining losses from this intra-trade pattern according to the definition of national water losses introduced by Chapagain et al. (2005). They lost about 18,000, 11,000, and 1,200 , respectively. This means that these three countries together are considered a main source of water supply in the Nile basin region. If they keep maintaining their export role, it may force them to protect their available water supply and compensate for the losses either by setting trade bans or by increasing the export price of the selected crops under consideration.

Fig. 3.

Water saved by crops' trade as compared with the water required to cultivate them domestically.

Fig. 3.

Water saved by crops' trade as compared with the water required to cultivate them domestically.

Only Kenya, Sudan, Congo, and Burundi can benefit significantly from the average pattern of trade flow. They save about 14,800, 6,900, 5,900, and 4,100 , respectively, while they need, on average, about 15,400, 2,600, 6,300, and 4,400 , respectively, to produce these crops domestically. While Kenya is classified as the biggest importer of green VW, the saved water amount in this country represents about 96% of the amount of water required to produce these crops domestically. Hence, the aim of VWT to save national water is satisfied in Kenya. This does not contradict with our finding that Kenya is one of the central exporters in the Nile basin. Its central position comes in terms of the number of links with its co-riparian neighbors, and does not come in terms of VW volume. It only exports about 5.6 to other Nile basin countries, and this amount is negligible compared with the exported volumes of green VW from the central exporters: Uganda, Tanzania, and Ethiopia. They export about 100, 72, and 53 , respectively. Thus, Kenya's central position does not affect its national green VW savings.

As previously mentioned in the sub-section ‘Selection of crops studied’, Egypt has almost zero green VWC, which means that it does not have values of green VWC for the selected crops. This explains why Egypt does not appear in Figure 3.

Identifying influential Nile basin countries in green VW export network

In order to identify the influence of each Nile basin country in VW exports, Figure 4 shows the out-degree centrality as calculated by Equation (3), and the eigenvector centrality as calculated by Equation (6). The two measures combined are used to identify the most influential exporting countries of green VW. Countries which have high out-degree and high eigenvector centrality scores simultaneously are identified as the most influential exporters. Accordingly, Kenya, Uganda, and Tanzania are the most influential and effective exporting countries in exchanging green VW of the selected crops. As mentioned in the sub-section ‘Properties of green VW export network in Nile basin countries’, Kenya has a non-vulnerable export structure as it imports large volumes of green VW from many exporters. This implies that Kenya is also an influential importer in the Nile basin region. It has a distinctive central position in the Nile basin and it may be a strong competitor in future water debates. Hence, the presence of Kenya, as well as Uganda and Tanzania is unavoidable in drawing trade policy and water management plans in the region.

Fig. 4.

Out-degree scores (a) and eigenvector centrality values (b). Both, combined, determine the most influential countries in the Nile basin.

Fig. 4.

Out-degree scores (a) and eigenvector centrality values (b). Both, combined, determine the most influential countries in the Nile basin.

Small world effect of green VW export network

For the yearly average green VW export network of Nile basin counties over the period 2000–2013, the average path length calculated by Equation (7) is 1.327, and the average clustering coefficient calculated by Equation (8) is 0.563. Hence, such a network has a somewhat large average path length and significantly high average clustering coefficient compared with a random graph with the same number of edges and the same node set of green VW export network. This is applied at 82% wiring probability. Consequently, the green VW export network of crops of the Nile basin satisfies the two properties of small world network. This implies that one country can reach others with a small number of steps, a positive feature that may facilitate any debates and resolve any potential conflict over water in the future.

Conclusions and policy implications

In this paper, SNA is used to analyze the yearly average of VW exchange structure for a group of selected crops cultivated in Nile basin countries over the period 2000–2013. The analysis employed degree, eigenvector centrality measures, and examined small world network properties. All Nile basin countries have a stable and balanced green VW export structure, and each country can absorb and cope with any potential food shocks and water crisis in any exporting country with which it is connected.

Cap-Net (2008) showed that the shared river basins give the sovereign countries the right to develop or control over resources located within their territory, especially water resources. This right has a negative impact on neighboring countries with high population growth rate as the accelerating population growth requires increasing water demand. According to FOA AQUASTAT7, Egypt, Sudan, and Eritrea have the highest dependency ratios (96.9%, 96.1%, and 61.7%, respectively) of the total renewable water resources flowing into them from neighboring countries. As a result, the water management in these three countries, as they want it to be, may lead to conflict with its co-riparian neighbors. However, green VW export network of crops in the Nile basin satisfies the two properties of small world network. This implies that one country can reach others with a small number of steps, and this may help in resolving any potential managerial conflict about water in the future.

Furthermore, all Nile basin countries follow an integrated water resource management (IWRM) approach to manage water resources usage in the region. According to Cap-Net (2008), IWRM is a process which promotes the coordinated development and management of water, land, and related resources, in order to maximize the resultant economic and social welfare in an equitable manner without compromising the sustainability of vital ecosystems. Obeng & Kóthay (2009) stated that the IWRM approach aims to (1) manage water resources across different sectors and scales (from local to international) and (2) establish good governance and regulatory frameworks. This means that all the different uses of water resources are supposed to be considered together, and take into account the overall social, economic, and environmental purposes.

Neubert (2008) mentioned the requirements for a successful VWT approach. These requirements are: (1) countries should have sufficient foreign exchange available to pay for crop imports, (2) countries should have an alternative strategy to employ rural workers who lose their jobs as a result of the reduction of irrigated areas, and (3) VWT approach should be subject to a good governance framework. These requirements achieve benefits for countries that adopt VWT approach in the form of national water savings which helps to overcome water shortage, and can avoid the need for infrastructure to increase water supply such as dams, water pipelines, inter-basin water transfer, etc.

If Nile basin countries choose VWT as a policy option, they should make sure that the strategic VWT is consistent with the approach of IWRM, as Horlemann & Neubert (2006) suggested. This consistency is satisfied if Nile basin countries import water-intensive crops to save national water resources. For instance, Kenya, according to what has previously been discussed, can follow VWT as a strategic solution for water resources management and face any potential water crisis in the future. Nevertheless, Nile basin countries may not have success implementing the VWT approach with IWRM framework if the region keeps maintaining only a small number of central or influential countries. The current application shows that only three countries are classified as the most influential green VW exporting countries in the Nile basin region: Kenya, Uganda, and Tanzania. The VWT approach assumes that the decision on restricting water use should be taken by central countries and enshrined institutionally and in law, but the IWRM framework assumes that water management should be increasingly decentralized and taken over by all the stakeholders. This potential incompatibility may be due to the nature of both the VWT approach and IWRM framework.

According to FAOSTAT, Egypt produced about 4 million tons of these crops on average over the period 2000–2013, and it only imports about 9,328 tons of these crops from co-riparian countries. This will impose pressure on the supply side of water resources and drive down food prices. As indicated by Zeitoun et al. (2010), Egypt imports VW of crops from the international market more than from the Nile basin countries. Then, the membership of Nile basin countries in the World Trade Organization (WTO) will help enable them to sustain international trade relationships and diversify their VWT partners.

This paper has confined itself to describing VW export network for green water in Nile basin countries. The case study used is conservative because it considered a sub-set of crops that consumes only 15% of green water. The main purpose of this work is to relate the approach of SNA with VWT, and to illustrate how it could be an effective policy implication tool in this direction. In order to get results of more practical value, our ongoing research will extend this paper in two dimensions: methodology and application. To enhance the methodology, temporal network analysis will be applied to drop the unrealistic assumption of flat averages over time. As for application, full coverage of water content will be considered; no matter if this water is green or blue, and more crops and livestock will be incorporated so as to increase representativeness.

Acknowledgments

We would like to express our gratitude and appreciation to the anonymous referees for their valuable and insightful comments and suggestions which considerably helped in improving the manuscript.

1

The ten Nile basin countries are: Burundi (BDI), Democratic Republic of the Congo (ZAR), Egypt (EGY), Eritrea (ERI), Ethiopia (ETH), Kenya (KEN), Rwanda (RWA), former Sudan (SDN), Tanzania (TZA), and Uganda (UGA). We used Sudan (former) as no data are available after its division. From now on, we will use Sudan instead of Sudan (former). ZAR is shortened to Congo.

2

See the sub-section ‘Selection of crops studied’ for the crops studied and why they are being selected.

6

In this context, connectivity refers to the number of links between nodes within the networked graph.

References

References
Adger
W. N.
, (
2006
).
Vulnerability
.
Global Environmental Change
16
(
3
),
268
281
.
doi:10.1016/j.gloenvcha.2006.02.006
.
Allan
T.
, (
1998
).
Watersheds and problemsheds: explaining the absence of armed conflict over water in the Middle East
.
Middle East Review of International Affairs
2
(
1
),
49
51
.
Cap-Net
(
2008
).
Conflict Resolution and Negotiation Skills for Integrated Water Resources Management
.
UNDP
,
New York
,
USA
.
Chapagain
A.
,
Hoekstra
A.
&
Savenije
H
, . (
2005
).
Saving Water Through Global Trade
.
Value of water research report series No. 17
.
UNESCO-IHE
,
Delft
,
The Netherlands
.
Chapagain
A.
,
Hoekstra
A.
&
Savenije
H.
, (
2006
).
Water saving through international trade of agricultural products
.
Hydrology and Earth System Sciences
10
(
3
),
455
468
.
doi:10.5194/hess-10-455-2006
.
Dalin
C.
,
Konar
M.
,
Hanasaki
N.
,
Rinaldo
A.
&
Rodriguez-Iturbe
I.
, (
2012
).
Evolution of the global virtual water trade network
.
Proceedings of the National Academy of Sciences
109
(
16
),
5989
5994
.
doi:10.1073/pnas.1203176109
.
Easley
D.
&
Kleinberg
J.
, (
2010
).
Networks, Crowds, and Markets: Reasoning About A Highly Connected World
.
Cambridge University Press
,
New York
,
USA
.
El-Sadek
A.
, (
2010
).
Virtual water trade as a solution for water scarcity in Egypt
.
Water Resources Management
24
(
11
),
2437
2448
.
doi:10.1007/s11269-009-9560-9
.
Estrada
E.
, (
2011
).
The Structure of Complex Networks: Theory and Applications
.
Oxford University Press
,
Oxford
,
UK
.
Fattah
M.
,
Zaghloul
A.
,
Ghaith
M.
&
El Ayoty
S.
, (
2016
).
Food gap optimization model using virtual water concept (case study: Nile basin countries)
. In
Nineteenth International Water Technology Conference
,
Sharm El Sheikh, Egypt
, pp.
19
.
Fraser
E. D.
,
Simelton
E.
,
Termansen
M.
,
Gosling
S. N.
&
South
A.
, (
2012
).
‘Vulnerability hotspots’: integrating socio-economic and hydrological models to identify where cereal production may decline in the future due to climate change induced drought
.
Agricultural and Forest Meteorology
170
,
195
205
.
doi:10.1016/j.agrformet.2012.04.008
.
Freeman
L. C.
, (
1977
).
A set of measures of centrality based on betweenness
.
Sociometry
40
(
1
),
35
41
.
doi:10.2307/3033543
.
Freeman
L. C.
, (
1978
).
Centrality in social networks conceptual clarification
.
Social Networks
1
(
3
),
215
239
.
doi:10.1016/0378-8733(78)90021-7
.
Hoekstra
A.
, (
2010
).
The Relation Between International Trade and Freshwater Scarcity
.
Economic Research and Statistics Division, World Trade Organization
,
Geneva
,
Switzerland
.
Hoekstra
A.
&
Chapagain
A.
, (
2008
).
Globalization of Water: Sharing the Planet's Freshwater Resources
.
Blackwell Publishing
,
Malden, MA
,
USA
.
Hoekstra
A.
&
Hung
P.
, (
2005
).
Globalisation of water resources: international virtual water flows in relation to crop trade
.
Global Environmental Change
15
(
1
),
45
56
.
doi:10.1016/j.gloenvcha.2004.06.004
.
Horlemann
L.
&
Neubert
S.
, (
2006
).
Virtual Water Trade – A Realistic Concept for Resolving the Water Crisis?
Deutsches Institut für Entwicklungspolitik gGmbH
,
Bonn
,
Germany
.
Kleinberg
J.
, (
2004
).
The small-world phenomenon and decentralized search
.
SIAM News
37
(
3
),
1
2
.
Kosorukoff
A.
, (
2011
).
Social Network Analysis: Theory and Applications
.
Cambridge University Press
,
New York
,
USA
.
Lee
S.
,
Mohtar
R. H.
,
Choi
J.
&
Yoo
S.
, (
2016
).
Analysis of the characteristics of the global virtual water trade network using degree and eigenvector centrality, with a focus on food and feed crops
.
Hydrology and Earth System Sciences
20
(
10
),
4223
4235
.
doi:10.5194/hess-20-4223-2016
.
Marchand
P.
,
Carr
J. A.
,
Dell'Angelo
J.
,
Fader
M.
,
Gephart
J. A.
,
Kummu
M.
,
Magliocca
N. R.
,
Porkka
M.
,
Puma
M. J.
,
Ratajczak
Z.
,
Rulli
M. C.
,
Seekell
D. A.
,
Suweis
S.
,
Tavoni
A.
&
D'Odorico
P.
, (
2016
).
Reserves and trade jointly determine exposure to food supply shocks
.
Environmental Research Letters
11
(
9
).
doi:10.1088/1748-9326/11/9/095009
.
Mohamed
N. N
, . (
2016
).
Productivity of rainfed agriculture of the upper Nile river
.
The Nile River.
Springer, Cham, Switzerland
, pp.
467
501
.
doi:10.1007/698_2016_108
.
Neubert
S.
, (
2008
).
Strategic virtual water trade – A critical analysis of the debate
.
Water Politics and Development Cooperation
123
145
.
doi:10.1007/978-3-540-76707-7_6
.
Newman
M. E.
, (
2010
).
Networks: An Introduction
.
Oxford University Press
,
Oxford
,
UK
.
Obeng
A.
&
Kóthay
L.
, (
2009
).
A Handbook for Integrated Water Resources Management in Basins
.
Elanders
,
Gothenburg
,
Sweden
.
Sartori
M.
&
Schiavo
S.
, (
2015
).
Connected we stand: a network perspective on trade and global food security
.
Food Policy
57
,
114
127
.
doi:10.1016/j.foodpol.2015.10.004
.
Shi
J.
,
Liu
J.
&
Pinter
L.
, (
2014
).
Recent evolution of China's virtual water trade: analysis of selected crops and considerations for policy
.
Hydrology and Earth System Sciences
18
(
4
),
1349
1357
.
doi:10.5194/hess-18-1349-2014
.
Shuval
H
, . (
2007
).
‘Virtual water’ in the water resource management of the arid Middle East
. In:
Water Resources in the Middle East
.
Shuval
H.
&
Dweik
H.
(eds),
Springer, Berlin, Heidelberg
, pp.
133
139
.
doi:10.1007/978-3-540-69509-7_12
.
Wasserman
S.
&
Faust
K.
, (
1994
).
Social Network Analysis: Methods and Applications
.
Cambridge University Press
,
Cambridge
,
UK
.
Zeitoun
M.
,
Allan
J.
&
Mohieldeen
Y.
, (
2010
).
Virtual water ‘flows’ of the Nile Basin, 1998–2004: a first approximation and implications for water security
.
Global Environmental Change
20
(
2
),
229
242
.
doi:10.1016/j.gloenvcha.2009.11.003
.