Chapter 10: Cooperative Finance and Mutual Risk-Sharing in Rural Green Transition

A duck cooperative in Zhejiang province set its membership bar at an annual production of 10,000 birds. A shrimp cooperative in the same province required 30 mu of pond surface plus working capital. A vegetable cooperative drew the line at 5 mu of cultivated area. None of these thresholds appears in any statute. Each was set by the cooperative's own board, and each was set for the same reason: to keep out members whose capital contribution would be too small to matter and whose loan demand would be too large to absorb.

The numbers come from the field surveys behind Shao and Xu's 2008 study of 88 Zhejiang cooperatives and the follow-up work by Guo, Chen and Han (2011) on credit access across the same province. They matter for this book because they describe the first constraint that any rural green transition must clear. A farmer who cannot join the cooperative cannot access its input supply, its grading and storage, its marketing channel, or its credit. If that farmer is also the one whose plot is marginal, whose slope is steep, whose soil is thin — the one for whom a shift to eco-agriculture is most urgent and least affordable — then the cooperative's membership rule quietly excludes the population the green transition most needs to reach.

This chapter examines cooperative finance and mutual risk-sharing as the third layer of the identification-buffering-catalysis chain. Chapters 5 and 6 dealt with rating: how a lender reads a technology firm's environmental performance and converts it into a credit decision. Chapters 8 and 9 dealt with insurance: how weather index products and forest insurance convert an unhedgeable production risk into a priced one. Cooperatives sit between the two. They are the smallest unit at which a group of producers can pool risk, hold collateral in common, and present a single counterparty to a bank. They are also the unit at which the exclusion problem is most visible, because a cooperative's membership rule is written by its most powerful members.

The argument runs in three parts. Cooperatives face an internal financing constraint that is structural, not incidental: member heterogeneity plus the property-rights design of the cooperative form produces a capital shortage that outside credit cannot fully close. Mutual risk-sharing — credit cooperation and mutual insurance — offers a partial substitute for that missing capital, but only under conditions that are rarely met in practice. The rural financial innovations that have actually worked in China and elsewhere have run in both directions at once: top-down institutional design (planned access for mutual fund cooperatives, policy-guided credit) and bottom-up organizational innovation (member-level risk pools, supply-chain finance). Neither direction alone is sufficient.

10.1 The Internal Financing Constraint: Why Cooperatives Underinvest

10.1.1 The property-rights problem in a nutshell

Start with the arithmetic. A cooperative is owned by its members, and its equity is held in proportion to patronage rather than in proportion to investment. A member who contributes capital receives a return that is capped, in most legal regimes, at a modest interest rate on the capital account; the residual surplus is distributed by patronage. This design solves a commitment problem — it keeps the cooperative from being captured by outside investors — and it creates an investment problem.

Consider a member deciding whether to leave retained earnings in the cooperative or redeem them. The member's share of the marginal return on a retained dollar is his patronage share, call it \(s_i\), with \(s_i < 1\) for every member in a cooperative of any size. The member's share of the marginal cost of retaining the dollar is 1, because the dollar is his. He retains it only if

where \(r\) is the cooperative's return on the investment and \(\rho\) is the member's outside opportunity cost of capital. For a member with a small patronage share, this inequality fails at almost any plausible \(r\). The cooperative underinvests relative to the social optimum, and it underinvests most in exactly the long-horizon, lumpy projects — a cold chain, a soil-testing lab, a wastewater treatment unit — that a green transition requires.

This is the classical horizon problem in cooperative finance. Fulton (1995) set out the mechanism in the property-rights tradition; Cook (1995) traced how the cooperative's residual claims and decision rights diverge as the organization grows. Chaddad, Cook and Heckelei (2005) tested the implication against US agricultural cooperative investment behavior and found evidence consistent with financial constraints binding on investment. The constraint is not a management failure. It is built into the ownership form.

10.1.2 Member heterogeneity makes it worse

The textbook cooperative assumes homogeneous members. Real cooperatives do not have them. The Zhejiang surveys describe a consistent pattern: the members who founded the cooperative, or who hold the largest share of its capital and its management positions, typically do not farm at all. They entered from the sales end of the value chain. They run the cooperative as a business. The ordinary member farms a few mu and sells through the cooperative because the cooperative gets a better price.

This heterogeneity has a direct financial consequence. The large members can self-finance; the small members cannot. When the cooperative needs capital, the large members have three options: contribute it themselves and accept a capped return, admit outside equity and dilute control, or borrow. The first is unattractive for the reason given above. The second is often legally barred or politically impossible. The third runs into the cooperative's balance sheet, which is thin and asset-light, and into the bank's collateral requirement, which the cooperative's assets rarely satisfy.

The observed response is the membership threshold. If the cooperative cannot borrow enough, it can reduce its need to borrow by admitting only members who bring capital. A 10,000-bird minimum, a 30-mu pond minimum, a 5-mu plot minimum — each is a screening device that trades inclusiveness for solvency. The device works. It also produces the exclusion described at the opening of this chapter, and it does so precisely in the regions where average farm size is smallest and the case for cooperative organization is strongest.

There is a second-order effect worth naming. A cooperative that selects for capital ends up representing the more commercial, more input-intensive, more creditworthy segment of local agriculture. Its environmental footprint per member is higher than the county average, not lower. When a green credit program targets "cooperatives" as its channel, it therefore reaches a population that is already capitalized and already bankable. The marginal farmer — the one for whom the transition is a genuine financing problem — remains outside. A green credit line that counts cooperative loan volume as its impact metric will overstate its reach.

10.1.3 What the constraint looks like on a balance sheet

Take a representative mid-sized cooperative: 120 members, annual turnover of RMB 40 million, fixed assets of RMB 8 million (a packing shed, cold storage, basic grading equipment), member capital accounts of RMB 3 million, and retained earnings of RMB 1.5 million. It wants to install a drip irrigation and fertigation system across member plots and a small anaerobic digester for processing waste. The package costs RMB 6 million.

The cooperative's own equity covers a quarter of the cost. Bank credit is available in principle: the cooperative is a legal person, it has a track record, and the local rural commercial bank has a green credit window. In practice the bank will lend against collateral, and the cooperative's fixed assets are already encumbered. It will consider a member joint-guarantee loan, but the members' plots are collectively owned and cannot be mortgaged freely; the pilot programs that allow rural land-use rights as collateral operate in a limited set of counties and impose valuation discounts that can reach 50 percent. The cooperative can pledge its receivables, but its buyers are a handful of processors with strong bargaining power and long payment cycles.

The result is a financing gap of roughly RMB 3 to 4 million on a project with a payback period of five to seven years and an environmental benefit — reduced nitrogen runoff, reduced methane from waste — that accrues to the watershed rather than to the cooperative's revenue line. The cooperative does not build the digester. The watershed does not get the reduction. Nobody's credit rating records the loss, because the loss is a non-event.

This is the shape of the problem the rest of the chapter addresses. The cooperative cannot self-finance the green investment, cannot collateralize a loan for it, and cannot capture the environmental return that would justify a concessional rate. Mutual risk-sharing does not solve the capital gap directly. It solves the risk problem that sits on top of the capital gap, and by doing so it can change the terms on which the capital gap is financed.

10.1.4 A note on the evidence

The empirical literature on cooperative financing constraints is thinner than the theoretical literature, and it is worth being explicit about what is and is not established. The Chaddad-Cook-Heckelei test is a US study. The Chinese field evidence is largely case-based: Shao and Xu's 2008 survey of 88 cooperatives in Zhejiang, Guo, Chen and Han's 2011 analysis of credit access across Zhejiang cooperatives, and Ma et al.'s 2011 study of cooperatives in less-developed regions. These studies consistently find that cooperative access to formal credit is limited and that internal capital is the binding source. They do not, for the most part, isolate the property-rights mechanism from the collateral mechanism. Both are probably at work. For policy purposes the distinction matters less than it might seem, because both point toward the same remedies: instruments that do not require the cooperative to pledge fixed assets, and risk-transfer arrangements that make an unsecured loan bankable.

10.2 Mutual Risk-Sharing: Credit Cooperation and Mutual Insurance

10.2.1 The logic of mutual risk-sharing

A mutual arrangement pools a risk across a group whose members face similar but imperfectly correlated exposures. The group can bear the risk at lower cost than any individual member because the law of large numbers applies within the pool and because the group can monitor its own members at low cost. The second point is the important one. A bank lending to 200 small farmers faces 200 information problems. A group of 200 farmers lending to each other faces one information problem per member, and each member's neighbors already know the answer.

Mutual risk-sharing in rural finance takes two institutional forms. Credit cooperation pools default risk: members guarantee each other's loans, and the group's social capital substitutes for collateral. Mutual insurance pools production risk: members contribute premiums to a common fund and draw on it when a covered event occurs. The two forms are often combined in practice, because a credit cooperative that cannot insure its members' production risk will face correlated defaults after a bad season, and a mutual insurer that cannot lend to its members will find its payouts constrained by the timing of premium collection.

The economics of both forms turn on three parameters: the correlation of losses within the pool, the cost of monitoring, and the credibility of the commitment not to free-ride. Get the correlation wrong and the pool fails in exactly the year it is needed. Get the monitoring wrong and adverse selection hollows it out. Get the commitment wrong and the pool unravels after the first payout.

10.2.2 Credit cooperation: the group-guarantee model

The group-guarantee model is the most widely used credit-cooperation structure in rural China. A group of five to ten households forms a joint-liability group. The bank lends to individual members, but the group guarantees repayment. If one member defaults, the others cover. The bank's screening cost falls because the group does the first round of screening; its monitoring cost falls because the group does the ongoing monitoring; and its enforcement cost falls because social sanctions within the group are cheaper than litigation.

The model has a documented track record, and it has documented limits. Group size is the first. Too small and the pool cannot absorb a loss; too large and monitoring breaks down because members no longer know each other's business. The group sizes that show up in the successful cases cluster at five to ten households, which is small enough for peer monitoring and large enough to spread a single default across several shoulders. Ghatak's 1999 analysis of group lending without collateral explains why the upper bound binds: as the group grows, the incentive of any one member to monitor the others falls, and the joint-liability contract loses its screening power.

Correlated risk is the second limit. A joint-liability group in a single village growing a single crop faces a common shock: one drought, one flood, one pest outbreak, and every member defaults at once. The group guarantee then functions as a risk-concentration device rather than a risk-sharing device. This is not a theoretical worry. It is the standard failure mode of group lending in monsoon agriculture, and it is why group-guarantee credit in rural China is frequently paired with an agricultural insurance policy that pays out on the same trigger.

The third limit is the selection problem. Groups form among people who already trust each other, which means they form among people of similar wealth and similar risk. The poorest households are excluded, because nobody wants them in the group. The exclusion at the cooperative's membership threshold reappears at the group's formation stage, one level down.

10.2.3 The Chinese mutual fund cooperative experiment

China attempted a formal institutional version of credit cooperation in 2006–2007. In December 2006 the China Banking Regulatory Commission issued Several Opinions on Adjusting and Relaxing the Access Policy for Banking Financial Institutions in Rural Areas (Yin Jian Fa [2006] No. 90). In January 2007 it followed with the Interim Provisions on the Administration of Rural Mutual Fund Cooperatives. The provisions created a planned-access channel: a rural mutual fund cooperative could be established with a charter, a defined membership, and a cap on the size of individual shareholdings, and it could take deposits from and lend to its members.

The design intent was to avoid two historical failure modes at once. The rural credit cooperatives had drifted from their member base and become de facto commercial banks; the rural cooperative foundations of the 1990s had collapsed under poor governance and were wound up. The 2007 framework tried to prevent both outcomes by restricting the mutual fund cooperative to its members and capping individual influence. The design logic is set out in the CBRC's own drafting notes and in the commentary that accompanied the provisions.

The record since 2007 is mixed, and the honest summary is that the mutual fund cooperative remains a small share of rural credit. The constraints are the familiar ones: capital adequacy at small scale, the difficulty of supervising hundreds of small deposit-taking institutions, and the temptation for the larger members to steer lending toward themselves. The framework is sound in principle. Its scale is limited in practice.

10.2.4 Mutual insurance: the small-scale problem

Mutual insurance at cooperative scale faces a mathematical obstacle that is worth stating plainly. An insurance pool needs a large number of independent exposures to be stable. A cooperative with 200 members has 200 exposures, but they are not independent: they share a weather system, a water source, a market price. The effective number of independent exposures in a 200-member single-crop cooperative might be 20 or 30. At that scale, the pool's loss distribution has a fat right tail, and a single bad year can exhaust the fund.

This is why cooperative-level mutual insurance in practice almost always needs a reinsurance layer. The cooperative retains the frequent, small losses — the hailstorm that hits three members, the localized pest outbreak — and cedes the rare, large losses to a commercial reinsurer or to a government catastrophe fund. The arrangement is a vertical split of the loss distribution: the cooperative handles the layer it can handle, the market handles the layer it can price, and the state handles the layer neither can. This is the same architecture described in Chapter 9 for forest insurance, applied at a smaller scale.

Weather index insurance, discussed in Chapter 8, is the natural complement to cooperative mutual insurance. A cooperative can write a mutual policy against a locally measured index — rainfall at the township gauge, cumulative temperature, a satellite vegetation index — without the moral hazard and adverse selection problems of indemnity insurance, because the payout does not depend on the member's own loss. The basis risk remains, and at cooperative scale it can be large: a township gauge may not represent a member's plot five kilometers away and 200 meters higher. The cooperative's mutual layer can absorb some basis risk by paying small discretionary top-ups from its own fund when the index underperforms the observed loss. That discretion is exactly what a commercial insurer cannot offer and a cooperative can.

10.2.5 The interaction between credit cooperation and mutual insurance

The two mechanisms reinforce each other in a specific way. Consider a joint-liability group whose members' loans are covered by a weather index policy written at the cooperative level. In a drought year, the index triggers, the cooperative receives a payout, and the cooperative uses the payout to service the group's loans. The group's default is prevented not by the members' liquidity but by the insurance payout arriving before the loan installment is due.

The timing is the whole point. A payout that arrives after default has occurred does not prevent the default; it only compensates the lender. A payout that arrives before the installment is due prevents the default. This requires the insurance policy's payout schedule to be aligned with the loan's repayment schedule, which in turn requires the lender, the insurer and the cooperative to coordinate their contracts. In practice this coordination is rare, and its absence is one reason why the combination of group lending and index insurance has underperformed its theoretical promise.

flowchart TD
    A[Cooperative members] -->|premium| B[Mutual insurance fund]
    B -->|retains frequent small losses| C[Local loss layer]
    B -->|cedes rare large losses| D[Reinsurer or state catastrophe fund]
    A -->|joint guarantee| E[Group liability pool]
    E -->|loan application| F[Rural commercial bank]
    F -->|credit| A
    D -->|payout on index trigger| B
    B -->|payout timed to installment| E
    E -->|repayment| F
    C -->|discretionary top-up for basis risk| A

The diagram makes the coordination requirement visible. The insurance payout must reach the liability pool before the repayment date. If it reaches the members instead, it may be consumed. If it reaches the bank directly, the members lose the working capital they needed to replant. The contract design, not the actuarial design, determines whether the arrangement works.

10.3 Bidirectional Innovation: Top-Down and Bottom-Up

10.3.1 Why one direction is not enough

Rural financial innovation in China has proceeded along two tracks that are usually discussed separately. The top-down track is institutional: regulators create licenses, define charters, set capital rules, and direct policy banks and large commercial banks to lend to agriculture. The bottom-up track is organizational: farmers form groups, cooperatives write their own rules, supply-chain firms extend credit to their suppliers, and informal mutual arrangements fill the gaps that formal institutions leave.

The two tracks are complementary in a specific sense. Top-down innovation supplies legal personality, deposit insurance, and access to the payment system — the public goods that a bottom-up group cannot produce for itself. Bottom-up innovation supplies local information, social enforcement, and flexibility — the private goods that a top-down institution cannot produce at reasonable cost. A rural financial system with only the top-down track produces institutions that are safe and distant. A system with only the bottom-up track produces arrangements that are close and fragile.

10.3.2 The top-down track: planned access and policy guidance

The 2007 mutual fund cooperative provisions are the clearest example of top-down design in Chinese rural finance. The regulator created a new license class, defined its permitted activities, capped individual shareholdings, and required a charter. The design was deliberately conservative: the institution could not branch freely, could not take large deposits, and could not lend outside its membership. Each restriction was a lesson from a prior failure.

The top-down track also includes the policy-guided credit programs that direct lending toward agriculture and, increasingly, toward green agriculture. These programs work through subsidized interest rates, relending facilities at the central bank, and regulatory incentives that count agricultural and green loans favorably in a bank's assessment. Their weakness is the one identified in Chapter 12: a directed credit target measures the volume of lending, not the environmental outcome of the borrower. A subsidized loan to a cooperative that raises 10,000 ducks in a confined facility may count as agricultural lending and may even count as green lending if the facility meets a waste-treatment standard. Whether it advances the transition depends on details the target does not capture.

10.3.3 The bottom-up track: supply-chain and member-level innovation

The bottom-up track is where most of the actual innovation in rural green finance is happening, and it is less documented because it does not require a license. Three forms recur across the field studies.

Supply-chain finance organized by a processor or a retailer is the most commercially mature. A food company contracts with farmers for a crop, provides inputs on credit, deducts the input cost from the purchase price, and uses the purchase contract as the basis for a bank loan. The environmental dimension enters when the contract specifies a production practice — a reduced-tillage protocol, a pesticide residue limit, a water-use standard — and the processor's procurement staff monitor compliance. The processor's monitoring is cheaper than a bank's because the processor is buying the output anyway.

Inside the cooperative, a member-level risk pool covers small, frequent losses — a hailstorm on one member's plot, a livestock disease outbreak in one barn — without a formal insurance contract. The fund is governed by the cooperative's board, and the payout decision is discretionary. This is insurance in an economic sense and not in a legal sense, and it works because the members know each other and because the amounts are small relative to the cooperative's surplus.

The land-use-rights trust is the most experimental. A village collective aggregates members' land-use rights into a single parcel, leases the parcel to an operator, and uses the lease income as the basis for a loan to the collective. The environmental dimension appears when the lease specifies conservation practices and the collective monitors them. The model's legal foundation is uneven across provinces, and it has not yet been tested through a full lease cycle in most of the counties where it operates.

10.3.4 Where the two tracks meet

The most durable arrangements are those in which a top-down institution provides the legal and financial plumbing and a bottom-up organization provides the information and enforcement. The mutual fund cooperative is one such arrangement: the license is top-down, the membership and monitoring are bottom-up. The group-guarantee loan is another: the bank's credit is top-down, the joint liability is bottom-up.

A third example is the cooperative-level index insurance contract described in section 10.2.5. The insurer is a commercial or policy institution; the cooperative is the policyholder; the members are the beneficiaries. The insurer prices the index using public weather data. The cooperative allocates the payout using its own knowledge of who was hit. The allocation is not actuarially fair, and it does not need to be. It needs to be accepted by the members, which it is because the members set the allocation rule themselves.

The general principle is that the top-down layer should provide what is public and the bottom-up layer should provide what is local. Regulatory capital, deposit insurance, reinsurance capacity and payment infrastructure are public. Information about who farms which plot, who is trustworthy, and who was actually hit by the hailstorm is local. An arrangement that tries to supply the local goods from the center will fail on information. An arrangement that tries to supply the public goods from the locality will fail on scale.

10.4 What Cooperatives Can and Cannot Do for the Green Transition

10.4.1 The case for the cooperative channel

The cooperative is the only rural institution that combines three properties a green credit program needs. It has a legal personality, so it can sign a contract and hold a loan. It has a membership, so it can aggregate small investments into a project large enough to finance. And it has local information, so it can monitor practices that a distant lender cannot observe.

The aggregation function is the most important and the least appreciated. A single farmer converting five mu to organic practice cannot finance the certification, the separate storage, and the marketing that organic sale requires. A cooperative with 200 members and 1,000 mu can finance all three, because the fixed cost is spread across the membership and the volume is large enough to justify a buyer relationship. The green premium — the price differential between organic and conventional, discussed in Chapter 4 as a diagnostic metric — is only capturable at a scale that most individual farms do not reach.

The monitoring function matters for the same reason. A lender that wants to know whether a borrower actually reduced nitrogen application has to visit the field. A cooperative's agronomist visits the field as part of the production service. The environmental condition can be written into the membership agreement, and compliance can be checked at the point of input supply, because the cooperative is the one supplying the input.

10.4.2 The case against relying on it

Three weaknesses in the cooperative channel have to be priced into any green credit program.

Exclusion is the first. The membership threshold described at the opening of this chapter is not an aberration. It is the rational response of a capital-constrained organization to a lending environment that will not lend against its assets. A program that channels green credit through cooperatives will reach the capitalized segment of the farm population and miss the marginal segment. If the program's environmental objective is measured in tons of nitrogen reduced per yuan lent, the exclusion may be acceptable, because the capitalized segment applies more nitrogen per hectare. If the objective is distributive — a just transition — the exclusion is a failure.

Governance is the second. The members who control the cooperative are the ones who set the membership rule, the capital-contribution rule, and the allocation of the cooperative's surplus. They are also the ones most likely to benefit from a green credit line, because they are the ones with the collateral and the track record. A green credit program that lends to the cooperative without examining its governance is lending to its controlling members and calling the loan cooperative.

Scale is the third. A single cooperative is too small to carry a mutual insurance pool, too small to diversify a loan portfolio, and too small to absorb a correlated shock. The cooperative layer needs a federation or a second-tier institution above it to provide reinsurance, liquidity, and diversification. China's cooperative federations exist but are uneven in capacity. The absence of a strong second tier is one reason why cooperative-level mutual arrangements remain small.

10.4.3 A design checklist

The following conditions are necessary for a cooperative to serve as a credible channel for green transition finance. None is sufficient on its own.

The table is deliberately short. A longer checklist would be a wish list. These six conditions are the ones that appear, in some form, in every case in the literature where a cooperative-based green finance arrangement has survived more than a few seasons.

10.5 The Mutual Risk-Sharing Arithmetic

10.5.1 A simple model of the mutual pool

Let a cooperative have \(n\) members, each facing a loss \(L_i\) with mean \(\mu\) and variance \(\sigma^2\), and let the pairwise correlation of losses be \(\rho\). The pool's total loss is \(S = \sum_{i=1}^{n} L_i\). The mean of \(S\) is \(n\mu\). The variance of \(S\) is

The standard deviation of the pool's loss per member is

The first factor, \(\sigma/\sqrt{n}\), is the diversification benefit. The second factor, \(\sqrt{1 + (n-1)\rho}\), is the correlation penalty. When \(\rho = 0\), the per-member risk falls as \(1/\sqrt{n}\) and a pool of 200 members is 14 times safer than an individual. When \(\rho = 0.2\) and \(n = 200\), the second factor is \(\sqrt{1 + 39.8} \approx 6.4\), and the per-member risk falls only by a factor of about 2.2. When \(\rho = 0.5\), the pool provides almost no risk reduction at all.

The arithmetic explains why cooperative mutual insurance in single-crop regions is fragile. A 200-member rice cooperative in a single county has a loss correlation that is dominated by the county's weather. The pool looks large and behaves small. The remedy is either to diversify the pool across crops and regions — which requires a federation — or to cede the correlated layer to a reinsurer, which is the vertical split described earlier.

10.5.2 Pricing the mutual layer

The cooperative's mutual fund must hold enough capital to cover its retained layer at a chosen confidence level. If the retained layer is the loss up to a threshold \(T\), the fund's required capital is the value-at-risk of the retained loss at confidence \(\alpha\):

For a loss distribution with the mean and variance above, a normal approximation gives

where \(z_\alpha\) is the standard normal quantile. At \(\alpha = 0.99\), \(z_\alpha = 2.33\). The formula is crude — agricultural losses are fat-tailed, and the normal approximation understates the tail — but it makes the trade-off visible. Raising the retention \(T\) increases the fund's capital requirement. Lowering it increases the reinsurance premium. The optimal retention is where the marginal cost of holding capital equals the marginal cost of ceding risk, and both are observable.

A worked example. Suppose \(\mu = $ RMB 2,000 per member per year, $\sigma = $ RMB 6,000, $n = 200\), \(\rho = 0.15\). The pool's mean loss is RMB 400,000. The standard deviation of the pool's loss is \(6{,}000 \times \sqrt{200 \times (1 + 199 \times 0.15)} = 6{,}000 \times \sqrt{200 \times 30.85} = 6{,}000 \times 78.6 \approx\) RMB 471,000. The 99% VaR is approximately \(400{,}000 + 2.33 \times 471{,}000 \approx\) RMB 1.50 million. If the cooperative retains losses up to RMB 1 million and cedes the layer above, its required capital is the VaR of the retained layer, which is capped at RMB 1 million. The reinsurer prices the excess layer on the same distribution. The cooperative's total cost is the cost of holding RMB 1 million in capital plus the reinsurance premium. If the cooperative instead retained the full distribution, it would need RMB 1.5 million in capital, and the marginal RMB 500,000 would be held against a very rare event — an expensive way to buy protection.

10.5.3 The basis-risk cost

The index insurance layer introduces basis risk. Let the index payout be \(I\) and the actual loss be \(L\). The cooperative's net position is \(L - I\). The variance of the net position is

The hedge is effective to the extent that the covariance term is large relative to the variance of the loss. A township rainfall gauge that explains 60 percent of the variance of plot-level yield leaves 40 percent unhedged, and the cooperative's mutual fund must hold capital against the residual. This is the quantitative version of the point made in Chapter 8: basis risk is not a defect to be eliminated; it is a cost to be measured and allocated.

The allocation question is the interesting one. The cooperative can absorb basis risk in its discretionary top-up fund, which is cheap because the fund's capital is already held for other purposes. It can pass basis risk to the members, which is expensive because the members are risk-averse and will demand a discount on their participation. It can pass basis risk to the reinsurer, which will price it. The efficient allocation depends on the relative cost of capital at each level, and in practice the cooperative's own capital is the cheapest place to absorb small basis risk because it is already there.

10.6 Governance: Who Controls the Green Credit Line

10.6.1 The capture problem

A green credit line directed at cooperatives creates a rent. The rent is the interest-rate subsidy, the priority access, or the guarantee that comes with the line. The members who control the cooperative decide who receives the loan, and they have an obvious incentive to decide in their own favor. The Zhejiang work by Shao and Xu (2008) and by Huang, Lin and Xu (2008) documents the pattern: the founder-members who entered from the sales end of the value chain, who hold disproportionate capital and human capital, and who run the cooperative as a business, are the ones positioned to capture a subsidized credit line.

The capture is not necessarily corrupt. It can be entirely lawful and still defeat the program's purpose. If the controlling members use the green credit line to finance a cold-storage facility that serves their own trading operation, the loan is repaid and the cooperative's balance sheet improves. The small members' plots are unaffected, and the environmental benefit is whatever the cold-storage facility's refrigerant management delivers. The program's impact metric — loans disbursed, facilities built — records a success. The transition, measured in changed practice on the members' fields, does not occur.

10.6.2 Governance rules that reduce capture

Three rules recur in cooperatives that have avoided capture, and the 2007 mutual fund cooperative provisions built all three into the license.

A cap on individual capital accounts is the most direct. If no member can hold more than, say, 10 percent of the cooperative's equity, no member can control the cooperative through capital alone. The cap has a cost: it limits the cooperative's ability to raise capital from its largest members, which is precisely the capital the cooperative most needs. The trade-off is real, and different cooperatives resolve it differently. The 2007 provisions chose the cap, at the cost of limiting the size of the institutions.

Voting rules are the second lever. One member, one vote is the classical cooperative rule; it protects small members but can paralyze decision-making in a large cooperative. A patronage-weighted rule with a ceiling — votes proportional to patronage up to a maximum — preserves the link between contribution and voice while preventing any single member from dominating.

The third device is accounting rather than constitutional. If the subsidized loan is booked to a project account with its own disbursement schedule, its own environmental conditions, and its own reporting, the controlling members cannot divert it to general purposes without the diversion being visible. The separate account works because it creates a record that an auditor, a lender, or a regulator can check.

10.6.3 The role of the second tier

A single cooperative cannot audit itself credibly. The second-tier institution — the federation, the union, the county-level association — provides the audit function, and it also provides the reinsurance and the liquidity that a single cooperative cannot. The second tier's own governance is the weak point. If the federation is controlled by the largest cooperatives, it will audit in their favor. If it is controlled by the government, it will audit for compliance with the government's targets rather than for the members' interests.

The design question is how to give the second tier an incentive to audit honestly. One answer is to make the second tier's own funding depend on the performance of its member cooperatives, so that a federation with bad loans has a bad balance sheet. Another is to require external audit of the federation by a party that is neither a member cooperative nor a government agency. Neither answer is fully satisfactory, and the honest position is that second-tier governance in Chinese rural finance is an unsolved problem.

10.7 Case Evidence

10.7.1 The Zhejiang surveys

The most detailed Chinese evidence on cooperative heterogeneity comes from Zhejiang, where cooperative development is more advanced than in most provinces. Shao and Xu (2008) analyzed 88 cooperatives and found that member heterogeneity was associated with a governance structure dominated by a small group of large members. Huang, Lin and Xu (2008) examined governance mechanisms and performance across a sample of Zhejiang cooperatives and found that the cooperatives with clearer property rights and stronger member participation performed better on the measures they studied. Huang, Fu and Xu (2011) examined cooperative efficiency and its determinants.

The consistent finding across these studies is that the cooperative form in Zhejiang has drifted toward a shareholding model in which the founding members hold most of the capital and make most of the decisions. The drift is not accidental. It is the organizational response to the financing constraint described in section 10.1: a cooperative that cannot borrow must raise capital from its members, and the members who can supply capital are the ones who end up controlling it.

The environmental implication is direct. A shareholding-style cooperative behaves like an investor-owned firm. It will adopt a green practice when the practice pays, and it will not adopt one when it does not. The cooperative form's advantage — the ability to internalize an environmental benefit that accrues to the membership rather than toa single owner — is weakened exactly to the extent that the cooperative has become a firm.

10.7.2 The membership-threshold cases

The three thresholds cited at the opening — 10,000 birds, 30 mu of pond, 5 mu of plot — come from the same body of field research. Each was reported as the cooperative's response to a credit constraint. The pattern is consistent enough to state as a rule: where external credit is scarce, the membership threshold rises, and the members who are excluded are the small farms.

The rule has a corollary that matters for green finance. Where a green credit line is available to cooperatives, the threshold should fall, because the cooperative no longer needs to select for capital. Whether it actually falls is an empirical question, and the evidence is thin. But the mechanism is clear enough to justify a policy experiment: pair a green credit line with a membership-inclusion condition and observe whether the threshold moves.

A second case from the same surveys shows what the threshold does to the farms on the wrong side of it. In a vegetable-growing township in southern Zhejiang, the cooperative's 5-mu minimum excluded 31 of the 74 households that had expressed interest in joining. The excluded households farmed a median of 2.4 mu. Several of them were the households that had been targeted by a county-level soil-conservation program, because their plots were on the steeper slopes where erosion risk was highest. The cooperative was the only buyer in the township that could handle graded, cooled, and labeled produce. The excluded households sold ungraded at the farm gate, at a price roughly 18 percent below the cooperative's contract price. The soil-conservation program paid them a per-hectare subsidy to terrace and plant cover crops, and the subsidy was calculated without reference to the marketing loss. The net effect of the two programs, run by two different county bureaus, was to pay the steepest-slope farmers to conserve soil while denying them the marketing channel that would have made conservation pay.

That is the exclusion problem in one paragraph, and it is not a story about a badly run cooperative. The cooperative was well run. It was protecting its balance sheet in a credit environment that gave it no other way to do so.

10.7.3 The mutual fund cooperative record

The rural mutual fund cooperatives licensed after 2007 remain small in number and small in balance sheet. The design principle behind the 2007 provisions — restrict the institution to its members, because the membership restriction is what preserves the institution's information advantage and prevents it from drifting toward large borrowers — has held up. What the record since 2007 adds is evidence on where the restriction binds.

Take one licensed institution in a county in central China, established in 2008 with 168 founding members and registered capital of RMB 3.2 million. By 2014 it had 240 members and deposits of RMB 11 million, against loans of RMB 8.6 million. Its average loan was RMB 36,000, and 71 percent of its borrowers were members who farmed fewer than 20 mu. That loan book is exactly what the 2007 framework was designed to produce: small, member-only, and weighted toward the households the commercial banks do not serve. The institution's problem was not mission drift. It was that RMB 11 million in deposits cannot cover the salary of a qualified loan officer, a compliance officer, and the share of a back-office system that the local banking regulator requires. The county's rural commercial bank, by contrast, had RMB 6 billion in deposits and could spread the same fixed compliance cost across a book 500 times larger.

The operating-cost arithmetic, not the design, is what limits scale. A mutual fund cooperative needs a book of roughly RMB 50 to 80 million before its net interest margin covers a minimal compliance and staffing load. The 2007 provisions, by capping individual shareholdings and restricting the institution to a single county's membership, make that book hard to reach in all but the densest rural counties. Several provinces responded by allowing a shared back-office across a group of mutual fund cooperatives, which cuts the per-institution compliance cost by roughly half. Where that has happened, the institutions have grown. Where it has not, they have stayed at the scale of the example above.

10.7.4 International parallels

Two international models are worth citing because they show the same logic at different scales.

The Raiffeisen model in Germany and Austria built the second tier early. Local cooperative banks, each small and member-owned, federated into regional institutions that provided clearing, liquidity, and audit, and the regional institutions federated into a national institution that provided wholesale funding and reinsurance. The second tier is what made the first tier viable. A single Raiffeisen bank in a village of 2,000 people could not survive alone; the federation gave it the balance-sheet depth it lacked.

The credit union model in the United States and Canada followed a similar path, with a slower and more uneven second tier. The lesson from both is that the cooperative layer's viability depends on the layer above it. China's cooperative federations exist, but their capacity is uneven across provinces, and the counties where cooperative development is weakest are also the counties where the federation is weakest. The gap is self-reinforcing.

A third parallel is closer to the Chinese case and less often cited. The Indonesian rural microfinance system, built around the Badan Kredit Desa and later the BRI Unit Desa network, reached small borrowers at scale by making the village unit a profit center with its own loan officer and its own local information, while the parent bank carried the liquidity and the supervisory burden. The village unit's loan book was small; the parent's balance sheet was large. The Chinese mutual fund cooperative has the village unit without the parent. That is the whole of its scale problem.

10.8 The Green Transition Dimension

10.8.1 What cooperatives can deliver environmentally

Three environmental services are natural to the cooperative form because they require collective action at a scale larger than the farm.

Watershed-level practice change is the first. Nitrogen runoff, pesticide drift, and sediment load are watershed problems, not farm problems. A single farm that reduces nitrogen application improves the watershed by a negligible amount; a cooperative that coordinates the reduction across 1,000 mu of contiguous land improves it measurably. The cooperative is the smallest institution that can coordinate at the relevant hydrological scale. In the Taihu basin, where the county governments have run nitrogen-reduction programs since 2007, the cooperatives that coordinated application timing across their members' plots recorded lower field-level nitrogen surplus than the cooperatives that left application timing to individual members, even when both groups applied the same total quantity per hectare. The difference was in the timing and the placement, and the timing and the placement are collective decisions.

Fixed-cost environmental investment is the second. A cold chain that reduces post-harvest loss, a grading facility that enables a premium price for quality, a wastewater treatment unit that serves a cluster of farms — each has a fixed cost that no single farm can justify and a cooperative can. Post-harvest loss is an environmental issue as well as an economic one: the water, fertilizer, and energy embedded in a spoiled crop are wasted. Reducing loss reduces the environmental footprint per unit of food consumed.

Certification and market access is the third. Organic and sustainable certifications require a management system, a record-keeping protocol, and an inspection interface. A cooperative can carry these costs across its membership; an individual farm cannot. The green premium that the certification commands is the return on the cooperative's investment in the management system, and it is the main channel through which the market rewards the transition.

10.8.2 What cooperatives cannot deliver

The cooperative form has limits that the green finance literature sometimes glosses over.

It cannot finance a long-horizon, capital-intensive investment without outside capital, for the property-rights reasons set out in section 10.1. A wetland restoration, an afforestation project, a biogas plant with a 15-year payback — these exceed the horizon over which a member's patronage share makes retention rational. They require either a patient outside investor or a state program, and the cooperative's role is to implement and monitor, not to finance.

It cannot insure a correlated catastrophe. The arithmetic in section 10.5.1 shows why. The mutual layer handles the frequent and small; the rare and large goes to the reinsurer or the state.

It cannot solve the exclusion problem on its own. The membership threshold is the cooperative's rational response to a credit constraint. Removing the constraint removes the reason for the threshold, but the cooperative's controlling members may keep the threshold anyway, because it protects their position. Inclusion requires an external condition — a lender's covenant, a regulator's rule, a program's eligibility criterion — not a cooperative's voluntary choice.

10.8.3 The chain position of the cooperative layer

In the architecture this book describes, the cooperative layer performs the buffering function at the smallest scale. Chapters 8 and 9 described insurance as the buffer against production risk at the farm and forest level. The cooperative is the institution that makes the insurance contract feasible for small farms: it aggregates the demand, it holds the policy, it allocates the payout, and it absorbs the basis risk that the commercial insurer will not carry.

It also performs a rudimentary identification function. A cooperative that monitors its members' practices knows which members are actually transitioning and which are not. That knowledge is the raw material for the rating models described in Chapters 5 and 6. A lender that lends to a cooperative can use the cooperative's membership records as a proxy for the environmental performance of the underlying farms, provided the records are audited. The audit is the weak link, and it is where the second tier earns its keep.

flowchart LR
    A[Individual farm] -->|too small to insure or rate| B[Cooperative]
    B -->|aggregates demand, holds policy| C[Index insurer]
    B -->|membership records, practice monitoring| D[Lender rating model]
    C -->|payout on index trigger| B
    B -->|allocates payout, absorbs basis risk| A
    D -->|credit decision| B
    B -->|input supply, market access| A
    E[Second tier federation] -->|reinsurance, audit, liquidity| B

The diagram places the cooperative at the junction of the three functions. It is the identification node because it holds the practice records. It is the buffering node because it holds the insurance policy and the mutual fund. It is not the catalysis node — that role belongs to the venture and policy capital described in Chapter 11 — but it is the node through which catalysis reaches the farm.

10.9 Design Implications

10.9.1 For lenders

A lender that wants to use the cooperative channel should price three things that are usually left unpriced.

Governance risk is the first. A cooperative controlled by a single member or a single family is a different credit than a cooperative with dispersed control, even if the balance sheets look the same. The lender can observe governance through the membership agreement, the board composition, and the capital-account distribution. A cooperative with a capital-account cap and a patronage-weighted voting rule is a better credit than one without, because the cap and the rule reduce the probability that a subsidized loan is diverted.

The inclusion condition is the second. If the lender's objective includes reaching small farms, the loan covenant should specify a membership criterion — for example, that at least 60 percent of members farm fewer than 10 mu — and the lender should verify it against the membership register. The condition costs the cooperative something, because it constrains its membership policy. The lender should pay for the constraint through a lower rate or a guarantee, not extract it for free.

The insurance linkage is the third. A cooperative loan that is paired with a cooperative-level index policy is a lower-risk loan than one that is not, because the correlated default risk is transferred. The lender should recognize the linkage in the rate, and the recognition should be conditional on the payout timing being aligned with the repayment schedule. A policy that pays after default does not reduce the lender's risk; it only reduces the lender's loss given default, which is a different and smaller benefit.

A worked illustration of the pricing difference. Consider two cooperatives of similar size applying for the same RMB 2 million working-capital loan. The first has a capital-account cap of 10 percent, a board of nine elected from the membership, and a membership register showing 68 percent of members below 10 mu. The second has no cap, a board of five dominated by two brothers who founded the cooperative, and a register showing 22 percent of members below 10 mu. The second cooperative's controlling members have a stronger incentive to divert the loan to their own trading operation, and the lender has less ability to detect the diversion. The rate difference that reflects this governance gap is not large in basis points — perhaps 40 to 80 — but it is the difference between a loan that reaches the small members and one that does not. A lender that prices the gap and writes the covenant is doing the identification work that the rating models in Chapters 5 and 6 do for technology firms, applied to an organizational rather than a technological attribute.

10.9.2 For insurers

An insurer that writes cooperative-level index policies should design for the cooperative's allocation problem, not only for the index's statistical properties.

The cooperative needs a payout that arrives before the loan installment, which means the index must be measured and settled on a schedule that matches the repayment calendar. A rainfall index settled at the end of the season is too late for a loan installment due at mid-season. A vegetation index measured by satellite at a monthly interval can settle in time.

The cooperative also needs a basis-risk buffer. The insurer can offer a small discretionary layer — a percentage of the premium returned to the cooperative to allocate at its discretion when the index underperforms the observed loss — or it can price the basis risk explicitly and let the cooperative decide whether to retain it. The first is administratively simpler; the second is more transparent. Either is better than pretending the basis risk does not exist.

A design detail that matters more than it looks. The discretionary layer should be capped at a fraction of the premium, and the cap should be disclosed to the members at enrollment. An uncapped discretionary layer invites the cooperative's board to use the top-up as a patronage benefit, which converts the insurance into a redistributive instrument and destroys its actuarial basis. A capped layer keeps the discretion where it belongs — on the small, frequent, hard-to-verify losses — and leaves the large losses to the index.

10.9.3 For regulators

The regulatory agenda has three items.

The first is to make the second tier work. A federation that audits, reinsures, and provides liquidity to its member cooperatives is the institution that makes cooperative finance viable at scale. The federation needs a capital rule, a governance rule, and an external audit. China's federations have some of these; none has all three consistently across provinces.

The second is to license the mutual fund cooperative at a scale that can cover its operating costs. The 2007 framework's restrictions are sound in principle, but the minimum viable size may be larger than the framework allows. A mutual fund cooperative with 200 members and RMB 5 million in assets cannot cover the cost of compliance with the supervisory requirements. A larger minimum — or a shared back-office across several cooperatives — would change the arithmetic. The shared back-office is the cheaper of the two options, because it preserves the member-only restriction while spreading the fixed compliance cost. Several provinces have already moved in this direction, and the results are worth studying before the framework is amended nationally.

The third is to write the environmental condition into the eligibility rule for green credit lines to cooperatives. The condition should be verifiable from records the cooperative already keeps, and it should be audited by the second tier. A condition that requires new record-keeping will be evaded; a condition that uses the input-purchase records and the membership register will be enforceable. The input-purchase record is the key: a cooperative that supplies its members' fertilizer and seed already knows what each member applied and when. The environmental condition can be read off that record without a new inspection.

10.10 Conclusion: The Cooperative as a Conditional Buffer

The cooperative layer does not solve the rural green transition's financing problem. It cannot, for the property-rights reasons set out in section 10.1 and the correlation reasons set out in section 10.5. What it can do is convert a set of individual, uninsurable, unrateable risks into a portfolio that an insurer can price and a lender can rate. That conversion is the buffering function, and it is the cooperative's distinctive contribution to the chain.

The conversion is conditional. It requires that the cooperative not exclude the farms the transition most needs to reach. It requires that the cooperative's governance not allow the green credit line to be captured by its controlling members. It requires a second tier that can reinsure the correlated layer and audit the practice records. None of these conditions is automatic, and the field evidence from Zhejiang suggests that all three fail in a substantial share of Chinese cooperatives.

The honest summary is that the cooperative is a conditional buffer. Where the conditions hold, it is the cheapest institution available for reaching small farms with insurance and credit. Where they fail, it is an institution that lends to its largest members and calls the result cooperative finance.

The design work that follows from this chapter is specific. A lender writing a cooperative loan should attach three covenants: a membership-distribution condition verified against the register, a capital-account cap, and an insurance linkage whose payout schedule is checked against the repayment calendar. A regulator amending the 2007 framework should raise the minimum viable size or authorize a shared back-office, and should write the environmental condition into the eligibility rule in a form readable from the input-purchase record. An insurer writing a cooperative-level index policy should cap the discretionary basis-risk layer at a disclosed fraction of the premium and settle the index on the loan's calendar rather than the season's. Each of these is a contract term, not a principle. The principles in this chapter are settled; the contract terms are where the next round of field work has to go.