Insight, Agriculture & Finance
To reach smallholder farmers, a development bank lent US$100 million to a commodity trader. That is not a shortcut. It is a diagnosis, and it revives an old question about what happens when credit, inputs and offtake all run through one relationship.
The Asian Development Bank has approved a US$100 million loan to Louis Dreyfus Company to support procurement, processing and export supply chains involving smallholder coffee and cotton farmers in Indonesia, Vietnam, India and Pakistan. The programme is designed to improve market access while connecting farmers with training, certified inputs, climate-adaptive practices and sustainability programmes, and its design explicitly addresses barriers faced by women farmers.
Notice the structure. To reach smallholders, the bank lent to the trader. Not to the farmers, not to a microfinance institution, not to a ministry of agriculture. To the company that already buys the crop.
That is worth reading as a diagnosis rather than as a shortcut, because it is one.
The default framing treats smallholder climate resilience as a technology-transfer question. Develop the drought-tolerant variety, demonstrate the water-saving practice, run the extension programme, achieve adoption.
The binding constraint is usually somewhere else. A farmer who cannot finance the input cannot buy the seed. A farmer who cannot sell predictably has no reason to invest in a crop whose price at harvest is unknown. And a practice that raises yield in expectation while raising variance in the short term is not adoptable by a household with no buffer, however well demonstrated it is.
Put simply, the capacity to adopt a resilience practice is a function of income predictability, and income predictability is a function of the buyer. Resilience sits downstream of the offtake relationship, and interventions that arrive without one are pushing on the wrong end of the chain.
This is the same shape of finding we set out about rice methane reduction, where the agronomy was settled and the adoption barriers were water control, risk tolerance and incentive. Different crop, different continent, same order of constraints.
So lending to the entity that provides the offtake, the inputs and the extension in one relationship is a reasonable response to the actual problem. It is also a structure that development economics has studied for forty years and has clear findings about.
When a single counterparty supplies credit and inputs and buys the output, the arrangement is what the literature calls an interlinked contract. Pranab Bardhan, and Braverman and Stiglitz in the early 1980s, established both why these arrangements are efficient and what they cost.
They are efficient for a specific reason. Smallholder lending fails on collateral and on enforcement. A farmer has no security to pledge and a lender has no cheap way to compel repayment. Interlinking solves both at once: the crop is the collateral, and the lender recovers at the point of purchase, so default requires the farmer to sell elsewhere, which is difficult and observable. The same relationship that carries the credit carries the sale, which collapses two expensive monitoring problems into one cheap one.
The cost is not usually exploitation in the crude sense. It is price opacity.
When inputs, credit and offtake are bundled into one relationship, the farmer receives a single set of terms and cannot decompose it. The interest rate is embedded in the input price, or in the price paid for the crop, or in both. There is no line item. A farmer can be entirely satisfied with the relationship, can be materially better off than before it, and can still be unable to state what the credit cost or to compare it with an alternative. Neither can the funder, unless somebody goes and reconstructs it.
The buyer supplies inputs on credit, provides training and certification, and purchases the harvest at settlement. Three transactions flow between the same two parties, and they are settled together as a single set of terms.
None of the three is separately observable to the farmer, or to a funder standing behind the transaction. The efficiency and the opacity come from the same feature: the bundling is what makes the credit possible, and it is also what makes its cost unobservable.
None of this is an argument against the loan, which may well be the most effective instrument available for the problem. It is an argument about what a funder in this structure should require as evidence, and it is a short list.
The realised farm-gate price, against the alternative. Not the contract price, the realised one after deductions, compared with what an unlinked farmer in the same district received in the same season.
The effective cost of the bundled credit. Reconstructable from input prices, advance timing and settlement terms. Somebody has to do the arithmetic, because the farmer cannot and the contract does not state it.
Whether the farmer can exit. The practical test of whether an interlinked relationship is a service or a constraint is what happens to a farmer who sells one season's crop to somebody else.
The explicit attention to barriers faced by women farmers is the right commitment, and the mechanism by which it succeeds or fails is specific enough to be designed for.
Interlinked contracts run through whoever holds the contracting relationship. That is normally a registered household head, and in most of these systems the registration is in a man's name. Training invitations go to the registered contact. Input allocations are issued against the registration. Payment is made to the registered account.
The consequence is well documented across the commercialisation of smallholder agriculture and it is uncomfortable. Formalising a supply chain can raise household income while reducing the share of that income women control, because it moves revenue from many small informal sales, often handled by women, into a single formal payment to a registered account, usually held by a man. Both effects are real and they run in opposite directions. A programme that reports household income and calls it a gender outcome has measured only the half that flatters it.
The design responses are not exotic. Register the person who does the work rather than the person who holds the title. Split payment options. Attendance quotas at training that are enforced rather than aspirational. And measurement at individual rather than household level, which is the only way to see the effect at all.
This is the layer where a programme like this is won or lost, and it is where our fieldwork consistently finds the largest gap between programme documentation and what has happened on the ground.
Separately, Convergence and the Australian government have opened a new cycle of the Asia Climate Solutions Design Grant, focused on climate-resilient agriculture and nature-based solutions in Southeast Asia. It funds teams designing financial vehicles rather than funding farms directly, with women's economic participation written into the mandate.
That is a correct read of the aggregation problem. Individual farmers are too small for institutional investors, unfamiliar technologies read as risky to banks, and the vehicle that bridges them has to be designed by somebody before it can be funded by anybody.
It also means the terms on which several thousand farmers will eventually be aggregated are being written now, by small design teams, on grant funding, years before any farmer sees a contract. The same is true of any shared arrangement many parties come to depend on, which is the argument we made about battery-swapping standards: the cheap moment to write the rules is before anyone is locked into them. That is an unusually high-leverage moment and it is almost entirely unobserved. Whether the resulting vehicle registers the farmer or the household, whether it makes the credit cost visible, whether it permits exit, and whether it pays the person who did the work are all decisions taken at design stage by people optimising for bankability.
The practical implication for anyone funding design work is to require that the design teams include somebody who has sat on the farmer's side of one of these negotiations. Not as a consultation exercise at the end. As a member of the team that drafts the terms.
Terms set at design time are cheap to influence and expensive to revisit. That is the whole argument.
The Lab works on this in regenerative agriculture and financial inclusion, and our work in Lombok sits on exactly this question of what a programme means to the household it lands on.
If you are designing one of these vehicles and want the farmer side understood before the terms are fixed, tell us what you need to know.
This is an independent insight piece by Transitions Lab. For the Lab's applied work, see Regenerative Agriculture & Land Systems and Financial Inclusion & Payment Systems. See also Equity Is the Wrong Money for a Warehouse on why matching the instrument to the shape of the cash flow is the same question at a different point in the chain. To discuss a study, see Contact.