The registry knows who the farmer is. The lender needs to know what she sold.

India had generated 91.9 million Farmer IDs by 19 March 2026, according to the Ministry of Agriculture and Farmers Welfare, each one linked to the national identity system and to land records. Kenya's KIAMIS has 6.8 million farmers on it. Zambia's electronic farmer registers, built with FAO's support, cover 4.3 million verified and traceable farmers, with more than a million transacting through the system each year. Nigeria is enrolling six million farmers biometrically. Ethiopia launched a registry in May 2025. Rwanda has 2.8 million farmers in its subsidy system and a full registry in development.

Nothing in the history of smallholder agriculture has counted farmers on this scale. The obvious question is what the counting has bought.

What the registries have measurably done

Read the evidence the registry builders publish about themselves and a pattern appears. India reports that cross-checking subsidy claims against the registry and Aadhaar recovered about 47 million dollars from ineligible beneficiaries, that Maharashtra used it to move disaster relief to 8.9 million farmers in five days, and that Chhattisgarh ran its paddy procurement for 3.2 million farmers through it in a single season. Kenya's registry underpins an e-voucher programme that the GSMA credits with a large expansion of fertiliser distribution between 2022 and 2024, and with a rise in maize production reported by the farmers registered on it. Zambia's registers give every farmer, cooperative and vendor a unique digital identity, and FAO describes transactions inside the system, from input redemption to crop sales, as digitally recorded and validated. Rwanda reports fewer duplicate records and faster subsidy payment.

These are real gains. Subsidy fraud, ghost beneficiaries, slow relief and poor targeting are old problems, and a verified registry linked to a national ID answers them directly. Anyone who has watched an input programme leak should be glad of it.

Now look for the finance results. The GSMA brief, which is sympathetic to registries and written to encourage more of them, offers one: in India, lenders can fetch verified farmer, land and crop details and approve working capital credit "in under 30 minutes, replacing what was previously a drawn-out, multiweek process." That is a claim about speed for a farmer a bank was already prepared to lend to. It is not a claim about a farmer who was previously unseen. For Nigeria and Ethiopia the finance benefits are listed as anticipated. For Kenya, credit management is described as a service that can be integrated, not one that has been measured. And the brief's sixth finding says plainly that government evaluation of registries remains limited on exactly this point: improved access to finance.

So the record, as it stands in 2026, reads: subsidies transformed, finance pending.

Why a subsidy and a loan ask different questions

The reason is not that the registries are badly built. It is that a subsidy and a loan need different information, and a registry was built by the party that pays subsidies.

A subsidy is a one-way transfer. To deliver it well you need to know three things: who the farmer is, where the farm is, and what is grown on it. That is exactly what a modern registry holds. India's system describes itself as three linked registries answering who the farmer is, where the land is, and what crops are sown. Once those are verified and linked to a national ID, fraud drops, targeting improves and money moves faster. The registry does the whole job.

A loan is a two-way promise. The lender's question is not only who you are but what you have done: what you sold, to whom, in what quantity, and whether you repaid the last time someone advanced you money. The GSMA brief has a precise term for this. It calls it an economic identity, and defines it as a profile that captures "transaction history," which for a farmer means "transactional data from the sale of agricultural products." Its own bottleneck analysis says that without "verified data on land tenure, yield history or transaction patterns," financial service providers cannot assess creditworthiness. The World Bank reached the same place in 2018: an official proof of identity is a prerequisite for financial services, but it is the starting point, not the credit decision.

Put the two side by side and the gap is visible. A registry answers who, where and what is grown. A lender also needs what was sold and what was repaid, across seasons. The first three are static facts that can be captured in a registration drive. The last two are events, and they have to be recorded at the moment they happen, by whoever is standing there when the crop changes hands.

Where the events go unrecorded

That is where the chain breaks, and the World Bank's Global Findex shows how widely. Across sub-Saharan Africa, nearly three-quarters of adults who receive payments for agricultural products are paid only in cash: more than 140 million people, 66 million of them women. A cash sale to a trader at the farm gate leaves no record in any registry, national or private. The event that would have proved the farmer's economic identity happens, and then it is gone.

This is why registration on its own cannot close the finance gap even in principle. The GSMA brief notes that 45 percent of adults in low and middle income countries cannot access credit within 30 days, largely for lack of a documented financial history. A registry documents the farmer. It does not, by itself, document the history, because the history is made in transactions the registry never sees.

Zambia's case is instructive precisely because its system goes further than most. FAO reports that transactions inside the programme, including crop sales, are digitally recorded. That is a real step toward economic identity for the farmers and sales that pass through it. It also shows the boundary: a sale is recorded when it happens inside a system that records sales. The cash sale outside any system, which Findex says is the majority case across the region, is not a failure of any particular registry. It is simply not an event any registry can see.

The cost that identity does not remove

There is a second reason finance lags, and it explains why even a perfect economic identity would not be the whole answer. A benchmarking study by Aceli Africa and Dalberg of 32,000 agricultural SME loans across five countries, Zambia among them, found that expected credit loss barely changes with loan size, while the cost of originating and servicing a loan consumes 16.4 percent of the smallest loans and under one percent of the largest. Better data attacks the assessment part of that cost: knowing who to lend to. It does much less to the servicing part: travel, monitoring, collection. A registry, and even a transaction record, can make a farmer visible to a lender. Neither can make her cheap to reach. That part is answered by whoever is already there.

Two objections worth taking seriously

The first is timing. India's registry scaled between 2024 and 2026. Kenya's is younger than its e-voucher. Lenders build products on new infrastructure slowly, and it is possible that the finance results are simply not in yet rather than absent. That is fair, and it is why the finding here is that finance is pending, not that it has failed. The honest test is whether, in two or three years, the registry builders can publish a credit-access result with the same confidence they now publish a subsidy result.

The second is that the constraint may be governance, not data. The GSMA brief's fifth finding is that lenders will not rely on a registry without transparent data-sharing and consent frameworks, and that custodians are reluctant to share with private actors. On this view the missing piece is the API and the law, not the transaction record. There is truth in it. But it does not remove the earlier problem. An open, well-governed registry that shares who, where and what is grown still hands the lender no repayment history. Both layers are needed. One is being built at scale. The other, for most smallholders, is not being built at all.

What follows

For a government, the implication is not to stop building registries; the subsidy results alone justify them. It is to recognise that a registry is the foundation for farmer finance, not the finance itself, and that the next layer is a record of trade that grows season by season. FAO's April 2025 standards for connecting farmer registries to social protection systems point in that direction. The equivalent standard for connecting registries to verified transaction records does not yet exist.

For a lender or a development finance institution, the implication is about where to look for the missing history. It will not be in the registry, however good. It will be wherever the crop is bought, because the buyer is the only party present at the moment the event that lenders care about takes place.

Our own reading, held as an interpretation rather than a proven result: the parties best placed to record a smallholder's economic identity are not governments or banks but the buyers who already stand at the farm gate, provided they record every purchase against a verified individual, keep the record accurately and with consent, and keep it for years. This is the layer DNC is building in the villages around Petauke, in Eastern Province: every purchase and every loan recorded against the individual farmer through Panjiia, so that a verified financial history exists where before there was only a cash sale and a memory. It is one layer among several, it is small, and on its own it is not the answer to a 140 million person problem. But it is the layer the registries, by design, leave to someone else.

A registry can say a farmer exists. Only a record of trade can say what she has done.