China’s pilot water-resource tax raised grain output by improving water quality and spurring green innovation, with the biggest gains outside core grain belts and where digital-economy development and environmental regulation are stronger.

Can water resource tax reform increase grain yield?—Evidence from China

Yashu Qin, Shiyao Yuan, Yue Wang, Luwei Wang · May 05, 2026 · Frontiers in Sustainable Food Systems

openalex quasi_experimental medium evidence 8/10 relevance DOI Source PDF

Using a staggered DID on Chinese prefecture-level data (2013–2019), the water resource fee-to-tax pilot increased grain yields, largely by cutting industrial wastewater discharge and stimulating regional green innovation that improved water-use efficiency.

Introduction Water plays an extremely important role in agricultural production. However, the impact of the water resource tax reform on grain yield has not yet been fully investigated. Methods Taking the pilot reform of water resource “fee-to-tax” as a quasi-natural experiment, we construct a panel dataset of Chinese prefecture-level cities from 2013 to 2019 and employ a multi-period difference-in-differences model, supplemented by double machine learning and a series of robustness tests, to identify the impact of the reform on grain yield. Results The results show that the water resource tax reform significantly increases grain yield. This finding remains robust after a battery of tests, including parallel trend tests, placebo tests, PSM-DID estimation, and the exclusion of special samples. Mechanism analysis indicates that the reform promotes grain yield mainly through two channels: reducing industrial pollution and stimulating green innovation. Specifically, the reform reduces industrial wastewater discharge, thereby improving agricultural production conditions, while also enhancing regional green innovation, which strengthens water-use efficiency and agricultural productivity. Heterogeneity analysis further shows that the grain-yield-enhancing effect is more pronounced in non-major grain-producing areas, regions with higher levels of digital economy development, and areas with stronger environmental regulation intensity. Discussion This study extends the evaluation of water resource tax reform from water-use and environmental performance to agricultural output, and provides empirical evidence for understanding the role of resource-based environmental taxation in promoting food security under rigid water constraints.

Summary

Main Finding

The pilot “fee-to-tax” water resource tax reform in China (phased rollout, 2016–2019) causally increased city-level grain output. The effect is robust to multiple checks and operates mainly through two mediating channels: (1) reduction in industrial wastewater discharge (improved irrigation/soil conditions) and (2) stimulation of regional green technological innovation (measured by green patents), which raises water-use efficiency and agricultural productivity. The positive effect is larger in non-major grain-producing regions, in areas with more advanced digital-economy development, and where environmental regulation intensity is higher.

Key Points

Policy evaluated: conversion of water resource fees into a volumetric water resource tax (pilot expansion 2016–2019; Resource Tax Law 2019).
Primary outcome: log of total annual grain output at the prefecture-city level (2013–2019).
Identification strategy: multi-period difference-in-differences (DID) using staggered pilot implementation; supplemented with double machine learning (DML) and a suite of robustness tests (parallel-trends test, placebo tests, PSM-DID, excluding special samples).
Robustness: results remain statistically significant after alternative specifications and tests.
Mechanisms:
- Pollution reduction: reform reduces industrial wastewater discharge → improves agricultural production conditions → higher grain yield.
- Green innovation: reform increases green patenting/innovation → better water-saving and cleaner production technologies → higher productivity.
Heterogeneity: larger yield gains where (a) the region is not a major grain producer, (b) digital economy development is higher, and (c) environmental regulation intensity (measured by environment-related word frequency in government reports) is stronger.

Data & Methods

Sample: Panel of Chinese prefecture-level cities, years 2013–2019.
Treatment coding: city-year indicator = 1 from the year the water resource tax pilot was implemented in that city onward; non-pilot years/cities = 0.
Main econometric model:
- Output_it = α + β·DID_it + γ·X_it + μ_i + λ_t + ε_it
- Fixed effects: city and year.
Controls included: regional GDP (log), urbanization rate, openness (FDI/GDP), industrial structure (secondary+tertiary share), environmental regulation intensity (text-based measure from city government reports), population density, fiscal pressure (expenditure/income), human capital (higher-ed enrollment share), among others.
Mechanism measures:
- Industrial pollution proxied by industrial wastewater discharge (China Environmental Statistical Yearbook).
- Green innovation proxied by green patent counts (CNRDS matched to WIPO green patent list).
Data sources: Ministry of Finance / State Administration of Taxation pilot notices; China City Statistical Yearbook; provincial/statistical yearbooks; China Environmental Statistical Yearbook; China National Research Data Service (CNRDS).
Additional methods: double machine learning to guard against high-dimensional confounding; propensity-score matched DID; placebo tests and parallel-trend checks.

Implications for AI Economics

Causal ML integration: The paper illustrates practical use of causal machine-learning tools (DML) alongside econometric DID—an approach AI economists should adopt for robust policy evaluation when many covariates or potential confounders exist.
NLP for policy measurement: Using text-analysis to quantify environmental regulation intensity showcases how AI/NLP can produce policy-relevant indices from government documents; similar methods can enrich empirical studies of regulation and institutional quality.
Digital-economy complementarity: Larger policy effects where the digital economy is stronger suggest complementarities between digital infrastructure/tech adoption (including AI-enabled precision agriculture, IoT irrigation control) and environmental taxation—AI investments can thus amplify returns to environmental policies.
Monitoring & enforcement: The effectiveness channel (pollution reduction + green innovation) points to roles for AI systems in monitoring water use and pollution (remote sensing, anomaly detection), optimizing enforcement, and targeting interventions to maximize agricultural outcomes.
Patent analytics & innovation measurement: Using green patent counts as a mediator highlights how AI-driven patent classification and trend detection can provide timely indicators of technological response to environmental policy.
Policy design lessons for AI-driven interventions: The staggered-adoption DID design and robustness strategy underline the need for careful temporal treatment when evaluating phased AI/technology rollouts (heterogeneous timing, spillovers, path dependence).
Caution on generalization: Effects are context-dependent (water endowments, regulatory capacity, digital infrastructure). AI-economics models and policy simulators should incorporate regional heterogeneity and institutional complementarities when projecting impacts of environmental or digital interventions on food security.

If you want, I can extract key econometric tables, reproduce the causal DAG in text form, or produce a short slide-ready summary highlighting figures and policy takeaways.

Assessment

Paper Typequasi_experimental Evidence Strengthmedium — The study uses a credible quasi-experimental design (staggered DID) and modern robustness tools (DML, PSM-DID, placebo, parallel trends) which support a causal interpretation, but potential threats remain: non-random selection into the pilot, possible contamination/spillovers between cities, staggered-DID weighting issues, measurement error in city-level grain yields and pollution, and only short- to medium-run post-treatment observation (2013–2019). Methods Rigorhigh — The authors implement a suite of contemporary econometric methods (multi-period DID, double machine learning, PSM-DID), conduct standard and extended robustness checks (parallel trends, placebo, exclusion tests), and perform mechanism and heterogeneity analyses, indicating careful empirical practice and attention to identification threats. SamplePanel of Chinese prefecture-level cities observed annually from 2013 to 2019, with variation in timing of adoption of the water resource fee-to-tax pilot; main outcome is city-level grain yield, with controls including environmental indicators (e.g., industrial wastewater discharge), measures of regional green innovation, and covariates for economic and institutional conditions (sample size not reported in text snippet). Themesproductivity governance innovation IdentificationMulti-period difference-in-differences exploiting staggered rollout of a pilot water resource "fee-to-tax" reform across Chinese prefecture-level cities (2013–2019), supplemented by double machine learning to flexibly control for confounders, propensity-score-matched DID, parallel-trends tests, placebo tests, and robustness checks. GeneralizabilityFindings are specific to China’s institutional and policy context (design and enforcement of the fee-to-tax reform) and may not generalize to other countries., Analysis is at the prefecture-city level; results may not translate to farm- or household-level impacts., Study covers 2013–2019 (short-to-medium run); long-run effects are unknown., Heterogeneous effects suggest limited applicability to major grain-producing regions or areas with different digital-economy or regulatory profiles., Potential spillovers between adjacent jurisdictions may limit external validity for non-contiguous policy implementations.

Claims (9)

Claim	Direction	Confidence	Outcome	Details
The water resource tax reform significantly increases grain yield. Firm Productivity	positive	high	grain yield	0.48
The reform reduces industrial wastewater discharge, which improves agricultural production conditions (mechanism linking the reform to higher grain yield). Other	negative	high	industrial wastewater discharge	0.48
The reform enhances regional green innovation, which contributes to higher grain yield by strengthening water-use efficiency and agricultural productivity. Innovation Output	positive	high	regional green innovation	0.48
The reform improves water-use efficiency (a channel through which it raises agricultural productivity). Firm Productivity	positive	medium	water-use efficiency	0.29
The grain-yield-enhancing effect of the water resource tax reform is more pronounced in non-major grain-producing areas. Firm Productivity	positive	high	grain yield	0.48
The grain-yield-enhancing effect is stronger in regions with higher levels of digital economy development. Firm Productivity	positive	high	grain yield	0.48
The grain-yield-enhancing effect is stronger in areas with stronger environmental regulation intensity. Firm Productivity	positive	high	grain yield	0.48
The main finding (that the reform increases grain yield) is robust to multiple checks, including parallel trend tests, placebo tests, propensity score matching DID (PSM-DID), and exclusion of special samples. Firm Productivity	positive	high	grain yield	0.48
Resource-based environmental taxation (the water resource tax reform) can play a role in promoting food security under rigid water constraints. Consumer Welfare	positive	medium	food security (via grain yield)	0.05