Farmland management right mortgage loan participation and agricultural return migration: evidence from Ningxia, China
Abstract
China’s farmland management right mortgage loan (FMRML) arrangement allows farmland management rights to serve as collateral for agricultural production finance, yet household-level evidence linking it to agricultural return migration remains limited. Using 2024 survey data from 532 rural households in FMRML implementation areas of Ningxia, this study examines agricultural return migration, defined as the return of a household member with migrant-employment experience who remained in the rural area for at least 3 months and engaged in agriculture or agriculture-related activities. In the fully adjusted Probit model, participation is associated with a 15.9-percentage-point higher predicted probability of agricultural return migration, conditional on household characteristics, land and agricultural operating foundations, and local financial-service conditions. The association persists when a demand-side proxy is used and across alternative model and control specifications, county/district and township fixed effects, propensity-score matching, and a heteroskedasticity-based instrumental-variable sensitivity analysis. The baseline coefficient also lies outside the random-label permutation distribution. Mechanism-consistent models associate participation with a larger net change in operated agricultural area per household member and a larger log change in long-term agricultural productive-asset stock; both components are positively associated with agricultural return migration conditional on participation and the baseline controls. Bootstrap inference for Karlson–Holm–Breen decompositions yields positive descriptive indirect associations for each component and jointly. Separate probability-scale comparisons indicate larger associations for trained than untrained households, for households operating contiguous rather than noncontiguous plots, and for households whose heads had at least six rather than fewer years of schooling; differences by terrain and contracted farmland area are not statistically discernible. The association is substantially larger among households embedded in socialized agricultural service organizations. These findings connect the financial mobilization of farmland management rights with household labor re-engagement and local agricultural operating conditions and suggest that household capabilities, plot configuration, and organizational services are important complementary conditions. Because the core variables share a common cross-sectional reference period, temporal ordering remains unresolved; the estimates represent conditional and mechanism-consistent associations. Policy design should coordinate agricultural finance with inclusive technical and organizational services, voluntary plot-adjustment support, and borrower safeguards.