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

Ecosystem overexploitation is a critical global environmental problem. Payment-for-ecosystem-services (PES) programs, like the Conservation Reserve Program in the US, Agri-Environmental Scheme in the EU, and Grain for Green Project in China, aim to address this by creating markets for conservation. However, few PES programs focus on grasslands and pastoralists. The Grassland Ecological Compensation Policy (GECP) in China is a unique example, being the world's largest PES grassland conservation program in terms of area, participants, and monetary transfers. Launched in two five-year phases (GECP-I and GECP-II), it aims to restore grassland ecosystems and raise herder income. Despite the substantial investment, few studies have comprehensively evaluated its impacts. Previous studies have yielded conflicting results regarding GECP's effect on grassland quality and livestock production, partly due to methodological limitations such as the lack of proper control groups. This study aims to fill this gap by providing a comprehensive evaluation of GECP's impact on grassland quality, herder income, and income equity, analyzing potential mechanisms, and exploring heterogeneous impacts across regions and socioeconomic groups.

The literature on PES programs is extensive, but studies focusing on grasslands and pastoralists are scarce. Landell-Mills & Porras (2002) provide a global review of markets for forest environmental services, while Adhikari and Agrawal (2013) review 26 PES programs, finding none specifically addressing grasslands. Studies on China's Grain for Green program offer some relevant insights (Cao et al., 2009; Chen et al., 2015; Deng et al., 2012; Uchida et al., 2009; Wu & Lin, 2010), but their focus differs from the pastoral context of GECP. Two existing studies on GECP, Hu et al. (2019) and Liu et al. (2018), provide limited insights due to their scope focusing on herder behavior and grassland quality, respectively, and methodological limitations. Börner et al. (2017) and Sims & Alix-Garcia (2017) offer broader perspectives on the effectiveness of PES programs and their challenges, which provide a crucial background for this study. The lack of comprehensive research on large-scale grassland PES programs motivates this study, aiming to fill this knowledge gap and contribute to the broader understanding of PES program impacts.

This study employs a mixed-methods approach combining remote sensing data and household survey data. For grassland quality assessment, the Normalized Difference Vegetation Index (NDVI) derived from MODIS satellite imagery was used. County-level data encompassing 630 GECP program counties and 386 non-program counties were analyzed using a difference-in-differences (DID) method to estimate GECP's impact on grassland quality. This DID analysis controlled for various factors like climate conditions and socioeconomic characteristics. An event study was conducted to test the parallel trends assumption. The household-level analysis involved a survey of 972 households across five pastoral provinces (Qinghai, Gansu, Inner Mongolia, Xinjiang, and Tibet). Data collected included household income, GECP payments, livestock production details, and various socioeconomic variables. Fixed-effects (FE) models were employed to analyze the impact of GECP on household income, disaggregating the effects on total household income, net pastoral income, non-pastoral income and non-program income. Further FE models examined the impact on income equity by income terciles and explored mechanisms and heterogeneous effects based on various factors like farm size, road density, grassland type, education level, and access to non-pastoral employment opportunities. The inverse hyperbolic sine transformation was used to handle zero values in some variables. Standard errors were clustered at the appropriate level to address autocorrelation.

The study reveals several key findings: 1. **Grassland Quality:** GECP led to a small but statistically significant improvement in overall grassland quality (measured by NDVI), approximately 3.2% increase at the county level and a small but statistically significant increase of 1.1% at the household level when payment intensity doubles. However, there was significant regional variation, and the impact varied by road density and pre-existing grassland quality. 2. **Herder Income:** GECP payments significantly increased total household income. A 10% increase in annual GECP payments led to a 3.66% increase in total household income per capita. However, the impact on specific income sources (pastoral, non-pastoral) was less significant. 3. **Income Inequality:** GECP exacerbated income inequality. Higher-income households, owning more land, received disproportionately larger payments, leading to a widening income gap between high, middle and low-income groups. 4. **Mechanisms:** GECP induced changes in herder behavior. While reducing sheep inventories, the impact on cattle was not significant. Herders increased supplementary feeding and rented additional grassland, suggesting a shift toward diversification and potentially mitigating grassland pressure indirectly. 5. **Heterogeneous Impacts:** The study identified various factors affecting GECP's effectiveness. Improved rural roads, larger farm sizes, and the presence of local grassroots measures or formal monitoring systems for grazing intensity positively influenced grassland quality improvements. Surprisingly, non-pastoral wages did not significantly impact the GECP's effect on grassland quality. Spatial heterogeneity was observed; GECP was more effective in grassland and meadow regions than desert regions, while the impact on non-pastoral income varied with education, road access, and per-capita grassland area. Specific statistical results are presented in Tables 1-7 within the paper, providing detailed coefficients, standard errors, R-squared values, and p-values.

This study's findings provide a nuanced understanding of GECP's impacts. The modest improvements in grassland quality highlight the complexities of ecological restoration and the need for complementary interventions. The significant income increase, coupled with exacerbated inequality, raises concerns about equity implications. The mechanisms identified reveal how herders adapt to policy changes, highlighting the importance of market development and diversification. The heterogeneity analysis underscores the need for flexible, locally tailored program designs. These results contrast with some previous findings, likely due to improved methodology, notably the inclusion of appropriate control groups. The study's findings offer valuable insights for policy adjustments, emphasizing the need to address income inequality and enhance program efficiency by strengthening monitoring and incentivizing sustainable grazing practices.

This comprehensive evaluation of China's GECP reveals a complex interplay between ecological and socioeconomic outcomes. While GECP successfully boosted herder income, its impact on grassland quality was modest, and it widened income inequality. The study suggests improvements by aligning GECP with PES principles (emphasizing voluntary participation and conditionality), investing in infrastructure (particularly rural roads), strengthening monitoring mechanisms (potentially through cost-effective informal systems), and supporting diversification into non-pastoral activities. Future research could focus on refining the program design based on these findings, exploring long-term ecological effects, and investigating the potential for alternative policy instruments to enhance both ecological and socioeconomic outcomes.

This study acknowledges several limitations. Income data relied on recall, potentially introducing measurement error. However, grassland quality data from remote sensing and GECP payments from bank records were less susceptible to such error. Robustness checks (RE models) were conducted to mitigate potential bias. The household survey's spatial coverage may not fully represent the entire grassland ecosystem, and the relatively short timeframe of the study may limit the capture of longer-term effects. The study primarily focuses on the five major pastoral provinces and may not fully represent the variations across all grassland regions in China.