Model Predicts Fiscal Limits on Performance-Linked Rural Cooperatives

A tripartite evolutionary game model calibrated to a grape cooperative in rural China predicts that paying rural cooperatives for verified services can encourage training, quality control and market access, but only within a local government’s fiscal capacity. When performance-linked payments become too expensive, the model predicts that officials will return to rewarding the creation of registered cooperatives rather than their performance, weakening the very service system the payments were intended to build. The study, published in Frontiers in Sustainable Food Systems, combines model-based strategic analysis with field evidence from Dingzhuang village in Jurong, Jiangsu.

The research distinguishes “creation support,” a basic operational subsidy paid to every registered cooperative regardless of its service level, from “function support,” a performance-linked subsidy provided when cooperatives deliver substantive services and support high-quality production. The model predicts that creation support can stabilize what the authors call symbolic compliance: an organization adopts the legal and organizational form of a cooperative, collects public funds and minimizes operating costs without providing meaningful services to members. Function support instead links public payments to observable performance, giving cooperatives an incentive to train farmers, enforce standards, coordinate inputs, build brands and connect producers with higher-value markets.

This distinction addresses a reported problem in China’s rapidly expanding cooperative sector. The study states that China had 2.14 million legally registered farmers’ cooperatives by October 2024, while the estimated share classified as hollow rose from 64.8% in 2014 to 81.9% in 2021. Those historical figures describe the reported pattern and are not causal estimates produced by the new analysis. The proposed mechanism is a principal-agent problem within China’s administrative system: central authorities set rural-development targets, but local governments receive part of their assessment through the number of cooperatives registered, potentially encouraging officials to count new entities rather than ensure that they work for farmers.

Dingzhuang is the successful case used to calibrate the model. From 1998 to 2014, grape cultivation expanded in the village while standards remained inconsistently applied and farmer income remained near CNY 21,000 per capita. In 2015, the Maoshan Town Party Committee helped establish the Dingzhuang Grape Cooperative Federation, bringing together 1,927 farmer households and 18,000 mu of vineyards, about 1,200 hectares, under seven member cooperatives and five family farms. Its “five unifications” covered varieties, technical guidance, inputs, quality standards, and branding and marketing. Reported per capita farmer income reached CNY 41,500 by 2020. The authors use these observations to parameterize the model; the before-and-after income difference does not by itself show that subsidies caused the increase.

The October 2024 calibration combined semi-structured interviews with 12 participants—three Federation managers, six farmers and three local officials—with field observations, administrative data, financial statements, subsidy documents and secondary production records. The interviews were recorded with consent, transcribed, anonymized and coded thematically by two researchers. Production costs, subsidies and premium income were measured directly, while preference weights and the probability of obtaining a premium through spot-market sales were estimated from interview responses and calibrated to reproduce the observed post-2015 transition. The researchers therefore describe the field work as model calibration, not a controlled evaluation of the subsidy policy or a population-level test of its effects.

Formalized as an evolutionary game, the model represents bounded-rational learning among three groups. Local governments choose between creation and function support; cooperatives choose between symbolic compliance and substantive service; and farmers choose between low-quality and high-quality production. Replicator dynamics trace the changing proportions of these strategies, while Jacobian eigenvalues and numerical simulations identify stable and unstable equilibria. Eight pure-strategy fixed points are algebraically possible, but each is stable only under specified parameter conditions. The simulations are conditional predictions under the model’s assumptions and calibration, rather than causal estimates of what subsidies will produce in other places.

One central threshold concerns whether higher quality is financially worthwhile for farmers. Upgrading can raise production costs, require new skills and leave farmers exposed to uncertainty over whether a premium will materialize. Under symbolic compliance, farmers bear the learning cost themselves, so they need a relatively large share of the quality premium to make the switch. Under substantive cooperative service, training absorbs the learning expense, while standardized grading, long-term arrangements and reputation mechanisms can make premium-bearing sales channels more dependable. In the Dingzhuang calibration, the minimum farmer premium share fell from approximately 0.42 under symbolic compliance to zero when cooperatives provided substantive service and the modeled reputation loss for non-compliance equaled the additional production cost.

This result makes the cooperative a potential bottleneck in the simulations. With function support, cooperatives shifted toward substantive service within about 0.4 simulated periods when their initial propensity was set at 0.8, whereas they did not shift under creation support. Farmers responded more slowly, adopting high-quality production after about nine periods in the function-support scenario and not upgrading under creation support. Reaching the model’s strongest equilibrium—function support, substantive service and high-quality production—also required the local government’s policy benefit and lower creation-support monitoring cost to exceed the additional monitoring costs and performance payments associated with function support.

The model also identifies the quality premium and its distribution as influential. Raising premium income accelerated convergence by all three participant groups, while increasing the farmer’s share of the premium could accelerate upgrading without directly expanding the government budget. In simulation, raising that share from 0.1 to 0.5 speedied farmer adoption, although a share of 0.9 slowed the cooperative’s move toward substantive service, indicating a trade-off in distributing returns. These results are conditional sensitivities within the simulation. They do not establish that a particular premium-sharing rate would generate the same response across regions or products.

More funding did not always produce a stronger equilibrium in the model. At a baseline performance-linked subsidy of CNY 300,000 per year, represented as 30 in units of CNY 10,000 per year, the simulation moved toward function support, substantive service and high-quality production. At 60, equivalent to CNY 600,000 per year, it predicted that the local government would reverse course and adopt creation support. The authors call this a “subsidy trap”: when grassroots expenditure becomes unsustainable without corresponding transfers from higher levels of government, officials favor the less expensive strategy of rewarding registration, and incentives for substantive service unravel. Simulations of six equilibrium scenarios found that low-quality traps could be stable and difficult to escape, while partial upgrading remained stable but vulnerable to a weak link among service provision, premium sharing and public incentives.

The authors recommend shifting payments toward independently verifiable outcomes, including farmers’ realized market-price premiums, third-party quality-test pass rates and verified income gains, rather than relying on self-reported training attendance or certification documents. They also propose phased implementation, higher-level transfers to protect local fiscal capacity, and context-specific standards for production-, service- and marketing-oriented cooperatives. The findings remain bounded by the study design. The model was calibrated on a successful case, the qualitative evidence came from only 12 interviewees, and political backing in Dingzhuang may have reduced coordination costs. The central conclusion is therefore a model-based conditional mechanism rather than a demonstrated universal effect: external support can produce substantive institutions when rewards track verifiable performance, but the pathway can collapse if fiscal pressure turns governments back toward paying for organizational existence.

Subject of Research: Evolutionary game analysis of subsidy design and rural cooperative governance in China

Article Title: From symbolic to substantive: How subsidy design shapes cooperative roles and high-quality rural industrial development?

Article References: Huang, M., Jiang, L., & Zhuang, J. (2026). From symbolic to substantive: How subsidy design shapes cooperative roles and high-quality rural industrial development?. Frontiers in Sustainable Food Systems, 10, Article 1949418. https://doi.org/10.3389/fsufs.2026.1949418

Image Credits: AI Generated

DOI: 10.3389/fsufs.2026.1949418

Keywords: rural cooperatives, performance-linked subsidies, agricultural policy, farmer behavior, quality premiums, rural industrial development, evolutionary game theory, China, symbolic compliance, substantive services, symbolic, substantive