Liu et al. (2019, Water Research) asked whether virtual water trade (VWT) can save water resources, tracing freshwater embodied in China's interprovincial trade through multiregional input-output (MRIO) tables for 2007, 2010, and 2012. They found that interprovincial trade raised national freshwater use (by 28.0, 13.6, and 20.3 billion m³/yr respectively), with value-chain-related trade the largest contributor, and that scarcity-weighted impacts grew less than use impacts. But the authors conceded that these estimates rest on sector-aggregated accounting that ignores within-sector process heterogeneity, a 2008 census water-use ratio frozen across study years, a withdrawal-based ('use') indicator rather than consumptive (evaporative) losses, and no propagation of data and model uncertainty into the water-savings metrics. The open problem is to determine, with quantified uncertainty, whether China's interprovincial virtual water trade raises or lowers national freshwater use and scarcity — the sign and magnitude of the national and scarcity-weighted trade-induced change in freshwater use (the balance of avoided water use, BAW, under the local self-production counterfactual) — under an accounting that simultaneously resolves within-sector process heterogeneity, uses time-varying sectoral water intensities, evaluates withdrawal and consumption metrics in parallel, and formally propagates uncertainty in input coefficients and water intensities into the results. Later work has addressed individual pieces (blue-water-consumption accounting — tracking the evaporative loss of surface and groundwater rather than withdrawals — with no-trade counterfactuals, crop-benchmark process analysis, multi-year dynamic studies, massively disaggregated prefecture water-use datasets), but no study integrates these refinements or attaches uncertainty bounds, and the trade-type shares reported by existing studies quantify different objects — gross transfer composition under a consumption metric versus attribution of the net rise in national water use under a withdrawal metric — so how far the water-savings conclusion depends on accounting conventions, and whether it is robust, remain undetermined.

Background

Water is embodied in traded goods: producing wheat, steel, or electricity anywhere consumes freshwater that is then virtually shipped to the consumer. This 'virtual water trade' (VWT) matters for China, where about two-thirds of water resources lie in the south while production and exports concentrate in the water-stressed north, and interprovincial trade in 2012 embodied roughly 30% of all national freshwater use. The standard tool for tracing such embodied flows is the multiregional input-output (MRIO) model — a system of regional economic accounts linked by trade, in which a Leontief inverse matrix converts final demand anywhere into production (and hence water use) in every region. A key policy question is whether trade saves water at the national scale: if goods are produced in provinces that use less water per unit of output than the importing provinces would, trade 'avoids' water use nationally. This is measured by the balance of avoided water use (BAW), defined in the order (virtual water actually embodied in the traded flows) minus (virtual water that importing regions would have used to produce the same goods locally), so that a positive BAW means trade raised national freshwater use (a net water cost) and a negative BAW means trade lowered it (a net water saving); and by the balance of virtual water embodied in trade (BVW); both can be weighted by local water scarcity, typically via a water stress index (ratio of water use to renewable availability), giving 'scarce water' variants. Liu et al. (2019, Water Research 163:114848) applied China's provincial MRIO tables for 2007, 2010, and 2012, disaggregating trade into three types: trade in final goods (TF), trade in intermediate goods for the final stage of production (TI), and value-chain-related trade (TVC, trade induced by the fragmentation of production across provinces). They found that interprovincial trade consistently raised national water use — by 28.0, 13.6, and 20.3 billion m³/yr in 2007, 2010, and 2012 — because water-intensive products increasingly flowed from less water-efficient to more water-efficient provinces only partly, with TVC contributing more than 37% of the increase; scarcity-weighted national impacts were smaller than the use impacts, since much of the embodied water came from provinces under lower water stress. The study's own Limitations section, however, concedes that these numbers are fragile: the MRIO sectors aggregate heterogeneous production processes across and within provinces; secondary-industry water-use shares rely on a 2008 census-based ratio applied unchanged to 2007, 2010, and 2012, so technological change in sectoral water intensity is ignored; the analysis measures freshwater 'use' (withdrawal) rather than 'consumption' (evaporative and other consumptive losses), which is the metric that depletes the resource; and none of these data and modeling uncertainties are propagated into the BAW/BVW or scarcity-weighted estimates, which are reported as point values. Subsequent studies have shown that such conventions are not innocuous: repeating the fragmentation analysis with a blue-water consumption metric (blue water being water drawn from surface and groundwater bodies, whose consumptive use is dominated by evaporation, in contrast to soil-moisture 'green water') and a no-trade counterfactual (Ye et al. 2022) still finds national worsening, but reports that direct final-goods trade contributed most of the gross volume of virtual blue water transferred (47% of the 2012 total) and value-chain-related trade least (17%) — whereas Liu et al.'s own gross trade-type shares of embodied trade were roughly equal across TI, TF, and TVC, and their finding that value-chain-related trade contributed more than 37% of the rise in national water use decomposes the net increase rather than gross transfers. These figures measure different objects, so they do not constitute opposite attributions of the same quantity — both studies agree that trade worsened the national balance — but they illustrate that the reported trade-type structure of the water-savings conclusion depends on accounting conventions that no study has yet evaluated jointly or under quantified uncertainty.

Problem Statement

Determine, with quantified uncertainty, whether China's interprovincial virtual water trade raises or lowers national freshwater use and water scarcity. Specifically: construct a virtual water trade accounting for China's interprovincial trade — across the trade types distinguished by the source study (final goods, intermediate goods for final production stages, and value-chain-related trade), and covering the source study's benchmark years 2007, 2010, and 2012 — that (1) differentiates production processes within sectors instead of aggregating them into broad multiregional input-output (MRIO) sectoral classes; (2) uses time-varying sectoral water-use intensities that reflect technological change rather than a single census-year (2008) ratio applied to all study years; (3) computes both freshwater withdrawal ('use') and consumptive ('consumption', i.e., evaporative and other depleting losses) indicators in parallel; and (4) formally propagates data and model uncertainty — at minimum in the input coefficients and the sectoral water intensities — into the provincial and national estimates. The required scientific outputs, jointly supporting this single objective, are: the sign and magnitude, reported with uncertainty bounds, of the trade-induced change in national freshwater use (expressed as with-trade minus counterfactual, so a positive value denotes a net national water cost) and its scarcity-weighted counterpart — equal to the balance of avoided water use (BAW) under the local self-production counterfactual and to the structural trade-versus-no-trade change in national freshwater use under the no-interprovincial-trade counterfactual — for every combination of the two water metrics (withdrawal 'use', consumption), the two sector resolutions (aggregated MRIO sectors, within-sector processes), the two intensity datasets (frozen 2008 census ratio, time-varying sectoral intensities), and the two counterfactuals (a per-flow local self-production counterfactual, in which each importing region substitutes local production at its own water intensity for the goods it imported, and a structural no-interprovincial-trade counterfactual computed as a standard MRIO autarky re-solve, in which each province meets its own final demand entirely from local production); a quantified sensitivity analysis identifying which accounting conventions (metric choice, sector resolution, intensity data, counterfactual) drive the differences between existing point estimates; and a defensible overall conclusion, with confidence levels, on whether interprovincial virtual water trade saves or costs China national water resources, together with the policy implications for which provinces, sectors, and trade types water-conservation measures should target.

References

  1. Xi Liu, Huibin Du, Zengkai Zhang, John C. Crittenden, Michael L. Lahr, Juan Moreno-Cruz, Dabo Guan, Zhifu Mi, Jian Zuo (2019). Can virtual water trade save water resources? Water Research 163:114848. DOI: 10.1016/j.watres.2019.07.015. https://doi.org/10.1016/j.watres.2019.07.015
  2. Quanliang Ye, Ranran Wang, Joep F. Schyns, La Zhuo, Lan Yang, Maarten S. Krol (2022). Effects of production fragmentation and inter-provincial trade on spatial blue water consumption and scarcity patterns in China. Journal of Cleaner Production 334:130186. DOI: 10.1016/j.jclepro.2021.130186. https://doi.org/10.1016/j.jclepro.2021.130186
  3. Jie Gao, La Zhuo, Yilin Liu, Pengxuan Xie, Wei Wang, Meng Li, Xuerui Gao, Pute Wu (2020). Efficiency and sustainability of inter-provincial crop-related virtual water transfers in China. Advances in Water Resources 138:103560. DOI: 10.1016/j.advwatres.2020.103560. https://doi.org/10.1016/j.advwatres.2020.103560
  4. Yueyue Du, Dandan Zhao, Meng Jiang, Yan Bo, Changxian Wu, Olli Varis, Jian Peng, Feng Zhou (2022). Local and non-local drivers of consumption-based water use in China during 2007–2015: Perspective of metacoupling. Journal of Environmental Management 312:114940. DOI: 10.1016/j.jenvman.2022.114940. https://doi.org/10.1016/j.jenvman.2022.114940
  5. Zhangqi Zhong, Zhuli Chen, Xiaojun Deng (2022). Dynamic change of inter-regional virtual water transfers in China: Driving factors and economic benefits. Water Resources and Economics 39:100203. DOI: 10.1016/j.wre.2022.100203. https://doi.org/10.1016/j.wre.2022.100203
  6. Feng Zhou, Yan Bo, Philippe Ciais, Patrice Dumas, Qiuhong Tang, Xuhui Wang, Junguo Liu, Chunmiao Zheng, Jan Polcher, Zun Yin, Matthieu Guimberteau, Shushi Peng, Catherine Ottle, Xining Zhao, Jianshi Zhao, Qian Tan, Lei Chen, Huizhong Shen, Hui Yang, Shilong Piao, Hao Wang, Yoshihide Wada (2020). Deceleration of China's human water use and its key drivers. Proceedings of the National Academy of Sciences 117(14):7702–7711. DOI: 10.1073/pnas.1909902117. https://doi.org/10.1073/pnas.1909902117
  7. Zongyong Zhang, Yuli Shan, Martin R. Tillotson, Philippe Ciais, Xu Zhao, Dandan Zhao, Hong Yang, Jingwen Huo, Zhao Zeng, Xian Li, Heran Zheng, Bofeng Cai, Wanqing Wang, Kai Wang, Guanlin Li, Geng Niu, Dabo Guan, Junguo Liu, Yu Hao (2025). Water-saving strategies across prefectures should target the manufacturing and agriculture sectors in China. Communications Earth & Environment 6:349. DOI: 10.1038/s43247-025-02292-3. https://doi.org/10.1038/s43247-025-02292-3
  8. Huiwen Liu, Huibin Du, Zengkai Zhang, Huimin Wang, Kunfu Zhu, Yaling Lu, Xi Liu (2023). Trade heterogeneity and virtual water exports of China. Economic Systems Research 35(3):397–416. DOI: 10.1080/09535314.2022.2035689. https://doi.org/10.1080/09535314.2022.2035689 (published online 2022)
  9. Yueyan Xu, Qingsong Tian, Yan Yu, Ming Li, Chongguang Li (2021). Water-Saving Efficiency and Inequality of Virtual Water Trade in China. Water 13(21):2994. DOI: 10.3390/w13212994. https://doi.org/10.3390/w13212994
  10. Jie Li, Zeyi Zhang, Diling Liang, Qianhong Ouyang, Peipei Tian, Dabo Guan, Heran Zheng (2026). China's Provincial Multi-Regional Input-Output Database for 2018 and 2020. Scientific Data 13:222. DOI: 10.1038/s41597-025-06543-y. https://doi.org/10.1038/s41597-025-06543-y
  11. Mohamed Badr, Konstantin Stadler, Juudit Ottelin (2026). A Comprehensive Review of Uncertainty in Multi-Regional Input–Output Modelling. SSRN preprint. DOI: 10.2139/ssrn.6352009. https://doi.org/10.2139/ssrn.6352009