Abstract
China's water resources are under increasing pressure from socioeconomic development, diet shifts, and climate change. Agriculture still concentrates most of the national water withdrawal. Moreover, a spatial mismatch in water and arable land availability--with abundant agricultural land and little water resources in the north--increases water scarcity and results in virtual water transfers from drier to wetter regions through agricultural trade. We use a general equilibrium welfare model and linear programming optimization to model interprovincial food trade in China. We combine these trade flows with province-level estimates of commodities' virtual water content to build China's domestic and foreign virtual water trade network. We observe large variations in agricultural water-use efficiency among provinces. In addition, some provinces particularly rely on irrigation vs. rainwater. We analyze the virtual water flow patterns and the corresponding water savings. We find that this interprovincial network is highly connected and the flow distribution is relatively homogeneous. A significant share of water flows is from international imports (20%), which are dominated by soy (93%). We find that China's domestic food trade is efficient in terms of rainwater but inefficient regarding irrigation, meaning that dry, irrigation-intensive provinces tend to export to wetter, less irrigation-intensive ones. Importantly, when incorporating foreign imports, China's soy trade switches from an inefficient system to a particularly efficient one for saving water resources (20 km(3)/y irrigation water savings, 41 km(3)/y total). Finally, we identify specific provinces (e.g., Inner Mongolia) and products (e.g., corn) that show high potential for irrigation productivity improvements.