Motivation
Over 90% of global water consumption occurs in agriculture, which depends on local rainfall, surface water, and groundwater. While water is costly to transport, agricultural trade can move water’s factor services indirectly by allowing water-intensive production to occur far from where crops are consumed.
Several prominent examples suggest that this process may contribute to regional depletion. Consider California’s Central Valley, where water tables have declined sharply as farmers specialize in water-intensive crops, like almonds, grown largely for export. Or consider northern India, where water depletion has outpaced nearly any other arable region, while farmers cultivate rice, of which India is now the world’s leading exporter.
As drought and groundwater depletion continue to become more salient, these examples raise a pressing question: Does agricultural trade worsen local scarcity, or does it preserve scarce water globally by shifting production toward places where water is relatively abundant?
Mechanism
The answer is not determined by standard gains-from-trade logic alone. Pervasive distortions affect both input markets for water and output markets for agriculture. Most farmers extract water under open access, without strong property rights or a market price, and agricultural markets face subsidies, taxes, tariffs, and trade restrictions. In such a second-best environment, trade can exacerbate depletion when it increases production where water is underpriced, a possibility emphasized by Chichilnisky (1994) and Brander and Taylor (1998).
Open access, however, does not make water scarcity irrelevant. Even without formal water prices, physical scarcity can raise farmers’ private costs. Where rainfall is limited, surface water is less accessible, or groundwater tables are deeper, farmers must devote more resources to obtaining water. This mapping from scarcity into extraction costs means that water’s effective input price can partly reflect local abundance, even when water markets do not operate.
We show that this mechanism is quantitatively important. Water-intensive agricultural activity is concentrated in water-abundant places (Fig. 1). The average water intensity of agricultural land use is much higher in regions with greater rainfall and groundwater availability. While parts of California and northern India are important exceptions, they stand out against a broader pattern in which drier regions mostly import water-intensive goods. In much of the Middle East and North Africa, for example, over 70% of agricultural water consumption is embedded in imports.
Fig.1
Counterfactuals
To evaluate this allocation, we use a dynamic spatial model of agricultural trade and water resources. The model allows crop production, trade flows, water extraction, and water tables to adjust over time. We compare the calibrated baseline with a counterfactual world in which countries cannot trade agricultural goods.
The autarky counterfactual shows that existing global trade substantially reduces aggregate land and water use. Without agricultural trade, global cropped acreage would be nearly double as countries meet domestic food demand using less suitable local land. Average crop yields would fall 44% after that reallocation. Global water extraction would rise 60.7%, and water tables would decline sharply over time as extraction exceeds recharge. Food prices would rise by a factor of 14.
The losses are concentrated in dry importing regions (Fig. 2). These regions rely on imports of water-intensive goods to avoid expanding local production. Under autarky, they bring substantially more land into agriculture and increase extraction where water is already costly to obtain. Thus, for many water-scarce regions, agricultural trade preserves long-run water availability.
Nevertheless, the model does not imply that trade reduces depletion everywhere. In a few rapidly depleting exporting regions, including California’s Central Valley and parts of northern India, autarky slows or even reverses declining water trends by reducing production of water-intensive crops for external markets.
Fig.2a-b
Policy
The same logic applies to agricultural policy reforms. The Uruguay Round reduced agricultural support in many water-abundant, high-income regions while reducing disincentives to agriculture in many poorer and drier regions. In our simulations, this shifted production toward some water-scarce locations, raised global extraction by 3.4%, and reduced initial global crop yields by 9.2%. By contrast, removing remaining agricultural distortions, which are largely subsidies, reduces global extraction by 3.7%.
The policy implication is therefore not that agricultural trade is unconditionally beneficial for water resources, nor that restricting trade is a general solution to depletion. In the presence of open-access water and distorted agricultural markets, the water consequences of trade policy depend on where production moves. Trade conserves water when it reallocates water-intensive agriculture toward places where water is relatively abundant. But it can worsen depletion when it reinforces production in places where water is already scarce.