Understanding government support for rural development in Hubei Province, China

The study investigates how government support for rural development can be identified and quantified at a geospatial scale by using legal rural construction land expansion (LRCLE) as a proxy in China, where the government exclusively allocates rural construction land. Rural development activities (e.g., agricultural production, rural infrastructure, public services) depend on construction land as a spatial carrier, yet prior work has not explicitly measured government support spatially. The research aims to construct and apply indicators capturing the intensity, spatial direction, and spatial agglomeration of government support via LRCLE in Hubei Province, and to determine which orientations of rural development (farmers’ production, rural public services, farmers’ life and social security, and industrial development) are being supported. Understanding these patterns is important for optimizing rural policies and aligning land allocation with rural revitalization goals.

The paper synthesizes research showing that multiple actors participate in rural development—government, social capital, villagers, and SMEs—with the government playing a catalytic and enabling role, including trust endorsement and provision of financing channels. Studies on governance structures emphasize the roles and power distribution between central and local governments and the need for empowering local governments to promote rural development. Policy-focused works highlight the importance of financial, agricultural, smart village, and related policies—typically government-led—in shaping rural trajectories. Literature on land use transitions underscores the bidirectional relationship between rural development and land use change, with rural construction land closely interacting with rural economic and social dynamics. Evaluation studies categorize rural development into farmland-, garden-, woodland-, and other natural resource-based forms, each generating demand for rural construction land (e.g., infrastructure, services, housing, processing, e-commerce, tourism). Prior empirical work shows heterogeneous correlations between socioeconomic factors and rural construction land expansion, implying selective allocation directions. The authors identify a gap: no prior research has quantified government support for rural development at geospatial scales via legal allocation data.

Study area: Hubei Province, central China, population 57.75 million (2020) with 21.43 million rural residents, large agricultural output, and 13 prefecture-level regions. Seventy county-level units outside urban areas were analyzed. Data sources: Spatial vector data of government-approved rural construction land (LRCLE) from 2009–2018 (patch-level attributes: approval ID, change date, area, location) from the Ministry of Natural Resources of Hubei Province; socioeconomic and resource-environment statistics from Hubei Statistical Yearbook (2010–2020), China Statistical Yearbook (county-level, 2010–2020), and provincial statistical bulletins (2009–2020). Indicator system:

• Characteristics of government support based on LRCLE patches include:
  1. Intensity of support: dynamic degree of single land use (DDSLU) calculated as the ratio of LRCLE area increment to base-period area over three subperiods (2009–2011, 2012–2014, 2015–2018).
  2. Spatial direction of support: standard deviation ellipse (SDE) of LRCLE using patch centroids weighted by patch area share to quantify centrality, dispersion (XStdDist, YStdDist), rotation, and ellipticity, tracking center-of-gravity shifts over time.
  3. Spatial agglomeration of support: local clustering assessed via 10 km × 10 km grid counts of LRCLE patches (DDLRCLE), class-level landscape metrics (patch density PD, edge density ED, largest patch index LPI, landscape shape index LSI, area-weighted fractal dimension FRAC_AM, and area-weighted Euclidean nearest-neighbor distance ENN_AM) to characterize structural and morphological changes.
• Orientations of rural development (independent variables) constructed across four dimensions: • Farmers’ life and social security: PPR (permanent resident population), PCDIRPR (per capita disposable income of rural permanent residents), ACL (area of cultivated land), TPAM (total power of agricultural machinery). • Rural public services: GBELG (general budget expenditure of local government), RIFA (rural investment in fixed assets). • Farmers’ production: TGO (total grain output), CPO (cotton production output), OPO (oilseed production output), TMPO (total meat production output). • Industrial development: GDPPI (GDP of primary industry), GDPSI (GDP of secondary industry), GDPTI (GDP of tertiary industry). Econometric analysis:
• Unit root tests (ADF) confirmed stationarity of dependent and independent variables; if needed, differencing would be applied (not necessary here).
• Multicollinearity tested by variance inflation factor (VIF), all below 10.
• Panel data regression models: Dependent variable 1: area of LRCLE per county-year (2009–2018); model selection via F test, Breusch-Pagan, and Hausman favored fixed effects (FE). Estimated one-way FE (individual and time) and two-way FE specifications. Dependent variable 2: number of LRCLE patches per county-year; model selection indicated pooled effects; pooled OLS estimated. Spatial analyses performed to compute DDSLU, SDE parameters, grid-based DDLRCLE distributions, and landscape indices across years.

• Intensity of support (DDSLU): Provincial averages increased across periods: 0.0147 (2009–2011), 0.0281 (2012–2014), 0.0573 (2015–2018), indicating rising government support intensity via LRCLE. Largest DDSLU by unit rose and shifted from the metropolis outward: 2009–2011 Wuhan highest (0.0749), Xiantao lowest (0.0000); 2012–2014 Ezhou highest (0.0449), Xiaogan lowest (0.0131); 2015–2018 Xiangyang highest (0.1057), Enshi lowest (0.0297).
• Spatial direction (SDE): The expansion scope (Shape-Area) was larger in 2012 and 2018, smallest in 2013. XStdDist and YStdDist generally increased; ellipticity declined overall (larger in 2009–2010, smaller in 2013). Rotation peaks occurred in 2011 and 2017; minimum in 2009. The center of gravity of LRCLE shifted northwest over time, evidencing clear directionality.
• Spatial agglomeration: Proportion of grid cells with zero LRCLE patches decreased from 47.32% (2009) to 28.83% (2018). Grids with DDLRCLE in ranges 1–5 and 6–10 increased then fell after 2016; 11–15 and 16–20 increased to 2017 then decreased in 2018; 21–25 and 26–30 fluctuated upward; 31–50 and 50–100 had small peaks in 2012 and overall fluctuating growth; cells with ≥100 decreased, with 2018 below 2010 levels. Landscape metrics showed: PD peaked in 2012 and stabilized 2014–2018; ED remained zero; LPI highest in 2009 then stabilized post-2014; LSI rose 2009–2012, dipped in 2013, then grew slowly; FRAC_AM stable with a maximum in 2013; ENN_AM highest in 2009 and stable afterward. These indicate increasing independence of patches, smaller patch sizes, and more complex shapes, suggesting more cautious and refined support under strict construction land controls.
• Model selection for area of LRCLE favored FE (F test P<0.01; Hausman P<0.01). For number of patches, pooled model selected.
• Determinants of LRCLE area (orientations of support): • Individual FE: Positive effects from PPR, GDPPI, OPO, ACL, TMPO; negative effects from GDPSI and CPO. Stronger effects for PPR and OPO. • Time FE: Positive effects from PCDIRPR, ACL, OPO; negative from CPO. Stronger effects for PCDIRPR and ACL. • Two-way FE: OPO, CPO, TMPO significant; OPO had the largest positive effect.
• Determinants of number of LRCLE patches (pooled regression): PPR, PCDIRPR, and TGO had negative effects; GDPPI and TMPO positive; ACL positive. PCDIRPR and GDPPI were significant drivers.
• Orientation summary: Government support via LRCLE primarily targeted farmers’ production (notably oilseed and meat outputs), followed by farmers’ livelihoods and social security; cotton output was associated with reduced LRCLE area.

The findings demonstrate that government support for rural development in Hubei, operationalized through legal rural construction land allocation, exhibits pronounced spatiotemporal patterns: rising intensity, clear directional shifts (northwest movement of the center of gravity), and increasing local agglomeration from single-center to multi-center patterns. Landscape metrics indicate growing patch independence and complexity, consistent with more cautious, fine-grained support amid strict controls on construction land growth. Econometric results clarify the orientations of support: allocations are more responsive to farmers’ production needs (especially oilseed and meat outputs) and to elements linked to food security and cultivated land protection, while cotton output correlates with reduced allocation area. The observed relationships between the number of patches and socioeconomic variables align with prior literature on drivers of rural construction land expansion (e.g., population and income relating inversely to patch counts, primary industry GDP positively associated). Policy context helps explain continued LRCLE expansion despite restrictive policies (e.g., urban-rural land linkages, basic farmland protection, quota systems), reflecting complex incentives where rural allocations can serve both rural needs and, indirectly, urban development pressures. Overall, the study addresses the research gap by providing quantifiable, spatially explicit measures of government support and identifying which rural development orientations are being prioritized.

• Government support intensity via legal rural construction land allocation increased from 2009 to 2018, initially concentrated in the metropolis and expanding to surrounding regions.
• Spatial directionality was evident, with a northwest shift of the expansion center and a transition from single to multiple centers of local agglomeration.
• Under stringent controls on construction land growth, support became more cautious, with smaller, more complex patches.
• Orientations of support differed by whether area or number of patches was considered. Not all support sought to increase area; in some cases, the demand was for more numerous, smaller allocations, contributing to spatial fragmentation. ACL and TMPO consistently required both area and quantity of LRCLE.
• Despite policies to curb rural construction land expansion and intensify land use, continued government-supported LRCLE reduces policy effectiveness, suggesting a need to recalibrate land quota and control mechanisms. Future research could extend the framework to other provinces, incorporate additional policy variables, and examine causal mechanisms linking specific policies to allocation patterns.