Abstract. Through the international trade of food commodities, countries virtually export or import the water used for food production, known as ‘virtual water’. The international trade network thus implies a network of virtual water flows from exporting to importing countries. The purpose of this study is to identify some controlling factors of the virtual water network by means of multivariate regression analyses, or gravity laws, as often named in economics. Starting from the FAOSTAT database, we reconstruct twenty-five years (1986−2010) of international virtual water trade values; we then analyze the dependence of the exchanged fluxes on: population, gross domestic product, arable land, virtual water embedded in agricultural production and dietary demand, and geographical distance between countries. Significant drivers are identified for each country considering separately export and import fluxes; temporal trends are outlined and the relative importance of drivers is assessed by commonality analysis. Results indicate that population, gross domestic product and geographical distance are the major drivers of virtual water fluxes, with a minor (nonnegligible) contribution given by the agricultural production of exporting countries. Such drivers have become relevant for an increasing number of countries throughout the years, with an increasing variance explained by the distance between countries and a decreasing role of the gross domestic product. The worldwide adjusted coefficient of determination of fitted gravitylaw model is 0.57 (in 2010), and it has increased in time, confirming the good descriptive capability of selected drivers for the virtual water trade.



Abstract. Trade theory suggests that a large number of factors influence production, consumption and trade of agricultural goods, and hence also the water content of the bilateral trade of agricultural products. This entails that to understand the sources of bilateral VWT ‘flows’ it is important to assess as many trade determinants as possible. The gravity model of trade, which relates trade flows to a number of monadic and dyadic features of the trading countries, stands as a very powerful tool to analyse the factors influencing bilateral VWT ‘flows’. Focusing on 145 developed and developing countries in the year 2006, this work estimates various specifications of a gravity model for the bilateral VWT ‘flows’ (associated with agricultural goods) and assesses the impact of water-related variables. Our findings show that the usual determinants of bilateral trade flows (such as geographical distance, economic mass, trade barriers, and the like) are statistically significant determinants of the water con- tent of bilateral trade in agricultural products. Water endowments and water pressure are found to matter as well: in a nutshell, countries tend not to over-exploit their domestic resources and, rather, they ac- quire the services of water through importing agricultural goods when domestic water is scarce and/or water pressure is high. These results follow economic intuition as one would expect that countries with scarce water resources tend to import the services of the water embodied in the incoming agricultural goods and to refrain from putting further pressure on their resources by specialising in ex- ports that are water-intensive. In fact, the literature on the impact of water endowments and pressure on VWT ‘flows’ has produced mixed evidence so far. Our analysis contributes to the literature as, differently from previous works, it addresses a large sample of countries, embeds alternative measures of water resources and pressure, and applies various estimation techniques, nonetheless delivering a consistent and robust message. The main conclusions in this work appear quite robust to changes in the estimation methods and to small variations in the composition of the samples under investigation. Major changes in the sample, however, can be conducive to results partially at variance with those found in the large sample. This suggests that the conclusions obtained from studies focusing on specific regions and countries should be generalized with great care. Moreover, this result helps to explain why many studies focusing on small groups of countries/regions reached conclusions that are at odds with the economic intuition tested (and confirmed) in this work.


Abstract The paper investigates the determinants of VW flows in the Mediterranean basin, testing the relevance of a large set of candidate variables and using a model averaging strategy that accounts for model uncertainty. Lacking a clear-cut theoretical framework, we draw on general economic theory and the existing literature on trade and on VW to select a number of potential determinants related to water flows: they range from classical water and land endowments to socio- economic variables such as trade policy and irrigation prices. Our results confirm one of the main controversial results in the literature: larger water endowments do not necessarily lead to larger ex- ports of VW as one could intuitively expect under the assumption that water availability may determine specialization in water-intensive agricultural activities. On the one hand, water is often not a binding constraint on agricultural output; on the other hand, there exist many competing uses of water (such as energy production and tourism), which may affect the relationship between water endowments and VW trade in a complex way. If water availability does not affect VW exports, then it is not surprising that we do not find a robust effect of water prices as well. Among the socio-economic variables, agricultural productivity and trade barriers appear to affect significantly VW trade. Having a better understanding of the main determinants of VW flows represents an important element in order to assess the potential benefits associated with those flows (for instance in terms of water saving) and to design policies aimed at improving the regional or global efficiency in the use of water resources. The paper suggests that the reduction in water availability in the Mediterranean, a likely result of climate change, may not disrupt the flows of VW. So long as VW imports allows countries to devote scarce water resources to other uses (which generate higher value added) or to relieve pressure on them, the resiliency of VW flows to water endowments can be viewed positively. Lowering trade barriers and improving agricultural productivity are two policy levers that can realistically be used to affect the magnitude of VW flows across countries.