Grassland ecological compensation policy in China improves grassland quality and increases herders' income

The study addresses whether and how China’s Grassland Ecological Compensation Policy (GECP), a large-scale PES program for grasslands, achieves its dual objectives of improving grassland quality and increasing herder income, and what distributional consequences arise. PES programs are increasingly used to address environmental degradation while supporting livelihoods, yet rigorous, counterfactual-based evaluations remain limited, especially for grasslands and pastoralists. Grasslands cover a substantial portion of global land, are ecologically and economically vital, and are widely degraded. China launched GECP-I (2011–2015) across eight provinces with substantial transfers to herders for reducing grazing, followed by GECP-II (from 2016) with expanded coverage and budget. Payments are proportional to certified grassland area and largely unconditional in practice. Prior studies on GECP are sparse and partial: Hu et al. examined herder behavior without assessing policy objectives; Liu et al. assessed grassland quality without proper control groups and did not study income. This study fills these gaps by jointly evaluating ecological and income outcomes, exploring mechanisms, and analyzing heterogeneity and equity impacts.

The literature on PES features prominent programs such as the US Conservation Reserve Program, the EU Agri-Environmental Scheme, and China’s Grain-for-Green. Despite growing adoption, counterfactual-based evaluations are relatively few, limiting generalizable conclusions. Reviews (e.g., Adhikari and Agrawal) note an absence of PES programs targeting grasslands and pastoralists. For GECP specifically, two studies exist: Hu et al. (2019) found limited effects on livestock production behavior in Inner Mongolia; Liu et al. (2018) reported large improvements in grassland quality but lacked proper control groups, potentially inflating estimates. Broader work underscores grassland degradation, their role in global meat supply, and the livelihoods of millions of pastoralists in China, motivating rigorous assessment of grassland-focused PES. This study advances the literature with a comprehensive, counterfactual-based evaluation linking ecological, income, and equity outcomes, and by investigating mechanisms and heterogeneity.

The evaluation leverages two panel datasets and corresponding econometric strategies. County-level panel data (2005–2015) cover 1,016 counties (630 GECP program counties in eight provinces; 386 non-program counties in five provinces). Grassland quality is measured by annual maximum NDVI derived from MODIS (MOD13A3, 1 km resolution). Controls include monthly climate variables (temperature, precipitation, PDSI) and socioeconomic variables (e.g., per-capita county fiscal income). County-level road network data from the Chinese Academy of Sciences are used to compute road density for heterogeneity analysis. A difference-in-differences (DID) model estimates GECP’s average treatment effect on NDVI, comparing treated counties (covered in GECP-I, 2011–2015) with control counties (not covered until 2016) before (2008–2010) and after (2011–2013) program onset. Event-study analyses test parallel trends. Robustness checks include alternative periods, groups, and random-effects models; standard errors are clustered at the county level. Mechanism analysis at county level examines impacts on log year-end cattle and sheep inventories. At the household level, a stratified random survey was conducted in five pastoral provinces (Qinghai, Gansu, Inner Mongolia, Xinjiang, Tibet) during 2017–2019 (with follow-up in 2020), sampling 972 households across 27 counties, 81 townships, and 162 villages. GPS coordinates of operated grasslands were matched to NDVI to construct farm-level grassland quality for pre- (2008–2010) and post-program (2015–2017) periods. Socioeconomic data include incomes (net pastoral, non-pastoral, GECP transfers), livestock inventories (converted to standardized sheep units), supplementary feeding, and grassland rental behavior. Payment intensity is defined as GECP payment per hectare; annual GECP payments are obtained from herder bank records for post-program years. Fixed-effects (FE) models with household and year fixed effects estimate impacts on: (i) farm-level NDVI as a function of log payment intensity; (ii) income outcomes (household income per capita, net pastoral income per capita, non-pastoral income per capita, non-GECP income per capita) as functions of annual GECP payment (inverse hyperbolic sine transformed). Controls include household-level (operated farm size, livestock labor, number of plots, joint-operated share, grass harvesting, crop/fodder planting), village-level (local grassroots measures limiting grazing, formal government monitoring, climate), and township-level variables (livestock prices, hay prices, non-pastoral wages, grassland rental prices). Standard errors are clustered at the village-by-year level. Event-study tests assess pre-trends for payment intensity exogeneity. Mechanism FE models analyze effects on livestock inventories (total, cattle, sheep), supplementary feeding, and rent-in behavior. Heterogeneity is examined at county level (by road density, initial NDVI terciles, and grassland type) and household level (by per-capita grassland area, presence of grassroots measures, formal monitoring, non-pastoral wage).

• Grassland quality (NDVI): Average NDVI increased 1.2% from pre- (2006–2010) to post-program (2011–2015) periods. DID estimates indicate GECP increased NDVI by 3.2% (county level). Household FE shows a 1% increase in payment intensity raises NDVI by 0.011% (doubling payment intensity ~+1.1% NDVI). Effects became positive roughly two years post-implementation.
• Income impacts: In Qinghai and Gansu, 2017 per-capita income was 13,136 RMB, with 28% from GECP (3,661 RMB), 56% pastoral (7,411 RMB), and 16% non-pastoral (2,064 RMB). In 2010, per-capita income was 9,100 RMB with ~90% pastoral and ~10% non-pastoral. Household FE: a 10% increase in annual GECP payment raises total household income per capita by 3.66%, with insignificant impacts on net pastoral, non-pastoral, and non-GECP income components on average (coefficients positive but not statistically significant).
• Income inequality: High-income households received much larger GECP transfers per capita than low-income households (7,440 vs 1,590 RMB; >4x). FE by terciles: a 1% increase in GECP payments increases income by 0.45% (low), 0.37% (middle), and 0.34% (high). Doubling GECP payments raises income per capita by ~407 RMB (low), 1,470 RMB (middle), and 6,955 RMB (high), widening absolute income gaps.
• Mechanisms: County-level DID shows GECP reduced year-end sheep inventories by 12.1% (p<0.05), with no significant effect on cattle inventories. Household FE indicates payment intensity does not significantly reduce total or cattle inventories; it increases supplementary feeding (10% higher payment per ha → +0.75%) and raises the likelihood of renting in grassland (10% higher payment per ha → +0.12 percentage points), suggesting low-level substitution and an incomplete rental market. Raising payment levels further, without other measures, is unlikely to reduce livestock inventories.
• Heterogeneity in ecological impacts: Counties with higher road density experienced larger NDVI gains (high road density +5.3%, medium +2.5%, low +1.7%). Medium-initial NDVI counties benefited most (+5.5%) compared to low (+3.5%) and high (+2.6%). By grassland type, positive and significant effects in grassland (+0.055), desert (+0.030), shrubland (+0.018), and herbosa (+0.038) counties; meadow positive but insignificant. At the household level, larger per-capita grassland area strengthens GECP impact (1% larger area → +0.016 percentage points to the payment effect); presence of local grassroots measures (≈+0.9% NDVI per 1% payment) or formal monitoring (≈+0.6%) enhances effectiveness; non-pastoral wage levels do not significantly moderate effects.
• Heterogeneity in income impacts: Greater improvements in household income per capita in grassland (≈+40.5%) and meadow (≈+23.2%) areas; insignificant in desert regions, reflecting higher payment standards in higher-quality grassland areas. Non-pastoral income gains are larger for households with more education, closer proximity to township-level roads, and smaller per-capita grassland areas.

The findings demonstrate that GECP achieves modest ecological improvements and significant income gains but increases income inequality among herders. The ecological benefits arise partly from reduced sheep inventories, alongside small increases in supplementary feeding and rent-in behavior, but additional payment increases alone are unlikely to further curb livestock pressure without complementary measures. Heterogeneity results underscore that enabling market access (roads), strengthening accountability (local grassroots measures or formal monitoring), and supporting larger, more efficiently managed farms can enhance ecological outcomes. The study also shows the importance of program design features grounded in PES theory. GECP’s largely unconditional, universal payments weaken incentives for behavioral change, explaining small average NDVI gains and the critical role of monitoring for effectiveness. Equity-efficiency tradeoffs are evident because land-based payments allocate more funds to land-rich (often higher-income) households, improving ecological efficiency but widening income gaps. Policy design should balance ecological outcomes, livelihood improvements, and equity by adjusting payment rules, incorporating conditionality, and enabling targeted support that expands non-pastoral opportunities and reduces reliance on grazing.

This study provides a comprehensive, counterfactual-based evaluation of China’s GECP, the world’s largest grassland-focused PES program. GECP modestly improves grassland quality and significantly increases herder income but intensifies income inequality, with high-income, land-rich households benefiting more in absolute terms. Mechanism evidence shows reductions in sheep inventories and small increases in supplementary feeding and grassland rental. Effectiveness is greater where road access is better, initial degradation is moderate, farm sizes are larger, and monitoring or local rules constrain grazing. Policy implications include enhancing conditionality and voluntary participation consistent with PES principles, investing in rural roads and human capital to facilitate non-pastoral employment, and leveraging cost-effective local monitoring to improve compliance. Future research should examine tailored strategies for severely degraded areas, the design and enforcement of conditionality, and improved measurement of mechanisms (e.g., supplementary feeding and land rental dynamics) at broader scales.

• Income data rely on respondent recall and may contain measurement error, potentially attenuating estimated income effects; robustness checks using random-effects models suggest consistency with fixed-effects results.
• Mechanism analyses are constrained by data limitations: lack of county-level measures for supplementary feeding and detailed grassland rental metrics limits cross-level validation.
• Monitoring effectiveness and compliance are imperfectly observed; informal and formal monitoring indicators capture presence, not intensity or quality.
• External validity may be influenced by GECP’s specific design (universal, largely unconditional payments) and China’s institutional context, which may differ from other countries’ PES implementations.