Skip to article content

Blockchain keeps being offered as a fix for how Indian farmers get paid. A 2022 review in the Journal of Agricultural Extension Management, published by MANAGE, the agricultural extension institute in Hyderabad, says the technology “allows widespread financial transactions between underutilized groups, without the need for intermediaries such as banks”. A small open-source project built on that review, farmer-payments-dlt, takes the claim seriously enough to write the code. It models three payment flows on a distributed ledger (DLT): escrow for crop sales, government scheme payments, and reconciliation against bank credits.

Its design note is unusually honest. It says a ledger is worth using only “where several parties who don’t fully trust each other need one shared, tamper-evident record”, that a single database should be used where one will do, and that money should keep moving on India’s existing rails (bank transfers, UPI and direct benefit transfer, or DBT). This piece takes that test and applies it to the places where farmer payments in India actually go wrong.

The short answer: most failures happen before any payment is made, in deciding who is eligible and which bank account belongs to them. A ledger records those decisions; it does not make them. The one place where several parties genuinely share a record they do not trust each other to keep, land records, is where an Indian state is now testing a blockchain.

Scheme payments: the problem is the list, not the transfer

PM-KISAN pays landholding farmer families ₹6,000 a year in three instalments of ₹2,000. The 24th instalment, released on 30 October 2026, sent about ₹18,690 crore to 9.34 crore farmers, taking the total since 2019 past ₹4.64 lakh crore. The transfer itself works: the agriculture ministry says the 19th instalment had a transaction success rate of 99.92%, which on that instalment’s 10.07 crore payees means roughly 81 thousand failed credits, retried later. It names the causes as Aadhaar numbers dropped from the National Payments Corporation of India (NPCI) mapper, which routes a payment from an Aadhaar number to one bank account, accounts not mapped at all, and closed accounts.

The larger numbers are elsewhere. The scheme began on trust, with states registering farmers on self-declaration and Aadhaar checks relaxed in some states. Income-tax payers, government employees and others not entitled to it were paid. The government’s DBT portal credits PM-KISAN with ₹22,106 crore of “savings” from deleting 2.12 crore ineligible names. Against money already paid to ineligible people, states had recovered ₹416.75 crore by December 2025. State audits show how hard recovery is: Bihar’s auditor found ₹62.67 crore recoverable and about ₹5 crore recovered by early 2022; Haryana’s found ₹40.70 crore paid to ineligible people and income-tax payers, of which ₹4.60 lakh had been recovered.

Tightening the list has also cut people off. Payees fell from 10.07 crore in the 19th instalment to 9.34 crore in the 24th, about 73 lakh fewer (7.2%). In March 2026 the ministry told the Rajya Sabha that 23.5 lakh beneficiaries had payments stopped because Aadhaar was not seeded to their bank account and 33.6 lakh because e-KYC was pending, alongside those whose land records were not yet seeded. If none of those overlap, that is up to 57 lakh farmers and about ₹1,141 crore withheld per instalment.

Look at what the project’s prototype does for scheme payments. Its disburse rule rejects a second payment for the same scheme, period and farmer, and rejects any farmer not marked verified. The duplicate check is a unique key, which any database enforces. The verification is a verified flag passed in by the paying agency itself: the ledger takes the agency’s word for it. That is not a flaw in the code; it is the point. Whether a farmer owns land, pays income tax or is the person holding the bank account is decided in land records, the income-tax database and the NPCI mapper. A ledger downstream of those would record ineligible payments just as permanently as eligible ones.

The government’s own answer is a better register. Under AgriStack, states have built farmer registries in which each Farmer ID is generated by matching a farmer to digitised land records; the count passed 11 crore on 23 October 2026, five months ahead of the March 2027 target, up from 10.28 crore on 30 July. The ministry says more than 6 crore farmers have already received scheme benefits through Farmer ID-based delivery. The register inherits the limits of the land records it is built on: only about 28,680 tenants and sharecroppers had been registered, in five states, by October.

Crop payments: escrow fixes a risk mills cannot afford to cover

The prototype’s second flow is an escrow. A buyer locks the price up front; the money is released when delivery is confirmed at the promised grade, and otherwise held until an arbiter decides the farmer’s share. This answers a real Indian problem: sugar mills paying for cane late.

The Sugarcane (Control) Order, 1966 requires mills to pay within 14 days of delivery, and states enforce it. In 2025-26 that rule held poorly. On 16 February 2026 mills owed farmers ₹16,087 crore, according to figures given to a parliamentary standing committee, with Karnataka (₹4,956 crore), Maharashtra (₹4,252 crore) and Uttar Pradesh (₹3,287 crore, across 121 mills) the largest. By 20 April, of ₹112,740 crore due for the season, ₹99,961 crore had been paid, leaving ₹12,779 crore. The US Department of Agriculture’s office in New Delhi put dues 97% cleared by 20 August, which leaves about ₹3,382 crore.

An escrow would have made those payments certain only if the mills had the money to lock up first. Mills are paying a cane price that has risen from ₹275 a quintal in 2019-20 to ₹365 for 2026-27, while the minimum price at which they may sell sugar has been held at ₹31 a kilo since February 2019. Arrears are a symptom of that squeeze. To escrow 14 days of purchases, mills would need roughly ₹8,769–10,522 crore set aside at any time, by this blog’s arithmetic, assuming the season’s ₹1.13 lakh crore of purchases is spread over a crushing season of 150–180 days. A bank would have to fund or guarantee that, and price it. The project’s own design note lists this as an open question: who funds an escrow if buyers pay on credit.

Mandi sales are a different case. e-NAM, the national electronic market, already pays sale proceeds directly into farmers’ bank accounts after an online auction. It had 1.89 crore farmers registered by June 2026 and has handled ₹4.84 lakh crore of trade since 2016. Its weakness is not payment but competition: an average of 2.8 trader bids per lot in 2025-26, and inter-state trade worth ₹14.28 crore that year and ₹83.95 crore in total, or 0.017% of all e-NAM trade. A tamper-evident log of bids would help show manipulation, which is what a 2020 study of three Uttar Pradesh mandis proposed after finding gaps between arrival prices and bidding prices. It would not add buyers.

Grade disputes have the same limit. In the prototype the buyer enters the delivered grade; if it is below the promised grade, the escrow goes to dispute. Whoever controls the grade entry controls the outcome. An independent assaying lab at the mandi solves that; a ledger only records the lab’s result.

Where each failure point lands

This blog’s assessment, using the project’s own test: a shared ledger helps where several parties who do not trust each other must keep one record. “Partly” means the ledger adds an audit trail, but the cause of the failure lies elsewhere.

Reconciliation: already being done, centrally

The third flow compares payments the ledger approved with credits the bank reports, and flags shortfalls. For government schemes that job belongs to the Public Financial Management System (PFMS), which issues each fund-transfer order, records the bank’s response and returns rejections with a reason code, such as Aadhaar not seeded in the NPCI mapper. A farmer can see the result through the PM-KISAN portal’s status page. The DBT Mission has told departments that account-number-based payments through PFMS will be switched off, leaving only payments routed by Aadhaar. The system is centralised, and the farmer cannot audit it, but the parties involved (the ministry, PFMS, NPCI and banks) are all regulated public or quasi-public bodies, which is a weak case for the multi-party distrust a blockchain is for.

Reading the prototype also shows how easy it is to leave a gap. Its reconcile function counts what a farmer is owed from released escrows and scheme disbursements only. When a disputed escrow is settled by the arbiter with a share for the farmer, that share is recorded on the ledger but never added to what is owed, so a bank’s failure to pay it would not be flagged. A test run here, with one disputed escrow settled at 60% of ₹10,000 and one ₹2,000 scheme payment, flagged only the ₹2,000. It is a small fix, and exactly the kind of issue a farmer could never spot from outside.

What the evidence elsewhere shows

The best-known farmer-payment case, Australia’s AgriDigital, settled its first grain sale on a blockchain in December 2016 and paid the grower within an hour. Even then, the payment ran through a digital wallet backed by conventional banking, and the company said it was not using blockchain for settlements during the extended pilot that followed. Index insurance pilots in Kenya and Sri Lanka used smart contracts to trigger payouts, but paid through mobile money. A 2026 review of such initiatives in Frontiers in Sustainable Food Systems puts the limit plainly: “Blockchain provides certainty about data integrity, not about counterparty intentions.” CGAP, the World Bank-hosted financial inclusion group, warned in 2018 that cost, complexity and context may erase blockchain’s advantages over conventional approaches.

In India, the project’s search found no blockchain payment pilot inside AgriStack, e-NAM or the crop insurance scheme PMFBY. Its survey of open-source code found agriculture repositories with between 0 and 34 stars on GitHub, mostly hackathon projects, none production-grade. The one live government deployment found for this piece is Andhra Pradesh’s Mee Bhoomi Blockchain, launched on 23 July 2026: land records from the revenue, survey and registration departments on Hyperledger Fabric, a permissioned blockchain, piloted on 1,37,763 parcels in 89 re-surveyed villages, with a state-wide roll-out planned from November. That is the right layer. Three departments keep separate records of the same land; disputes over which one is right are what block mutation, loans and, downstream, scheme payments. Whether it works will show in mutation times and dispute counts, which have not yet been published.

The prototype, briefly

The farmer-payments-dlt code is a teaching model of about 140 lines of standard-library Python with seven passing tests. Each block is hash-chained to the previous one and carries a keyed-hash (HMAC) signature from the consortium member who added it, so tampering is detectable and only members can write. It has no network, no consensus between nodes, and no public-key signatures; escrow states and the duplicate register live in memory rather than being rebuilt from the chain. It records instructions and confirmations, never money. As its README says, a real system would run on a permissioned platform such as Hyperledger Fabric, or on an Ethereum-compatible chain with audited contracts. Its value is the documentation: the reading of the 2022 review, the list of challenges and how a consortium design answers each, and the plain statement of what stays off the ledger.

What would fix this

• Publish PM-KISAN’s ineligible-payment ledger, state by state. The government publishes the amount recovered (₹416.75 crore) but not, in its recent releases, the total paid to people later found ineligible. Without that figure the recovery rate cannot be judged.
• Publish stop-payment and failure counts with every instalment: how many farmers were held for e-KYC, bank seeding or land seeding, how many credits failed, and how many were later paid, so the fall from 10.07 crore to 9.34 crore can be split into cleaning and exclusion.
• Seed the bank account at enrolment. The DBT Mission already asks schemes to check NPCI mapper status when Aadhaar is authenticated. Making that check part of Farmer ID creation would catch failed payments before they happen.
• Price cane escrow before mandating it. A bank-guaranteed escrow for cane, backed by a mill’s sugar or ethanol receivables, could be tested in one state with high arrears. Enforcing the interest that the Sugarcane (Control) Order already allows on late payment would come first.
• Put assaying labs, not ledgers, at the centre of e-NAM grade disputes, and log results so they cannot be edited afterwards; that log is where a hash chain is cheap and useful.
• Measure Andhra Pradesh’s blockchain land records against its old system on mutation time, disputes and fraud before other states copy it.

These are proposals from this blog, not recommendations any agency has adopted.

What this piece could not establish

The total amount paid to ineligible PM-KISAN beneficiaries nationally was not found in a recent official source; a figure of about ₹4,350 crore circulates but could not be traced to a Parliament answer or audit report, so it is not used. In 2021 the government was asked about recovering about ₹3,000 crore from 42 lakh ineligible beneficiaries, but that figure appears in the question, not in a confirmed answer. The count of beneficiaries stopped for pending land seeding in March 2026 was not read. The stop-payment categories may overlap, so the 57 lakh total is an upper bound. The February cane arrears and the comparison with earlier years come from press reports of a standing committee report; one report said arrears were up about ₹2,049 crore on the previous season, another said the previous figure was ₹497 crore, and the earlier-year comparison is therefore not used. The August cane figure is calculated from a rounded 97%. The escrow sizing assumes a 150–180 day crushing season and an even pace of purchases. No cost per payment, time-to-credit data or outcome study was found for any blockchain farmer-payment pilot, in India or abroad, and no results have been published for Andhra Pradesh’s land-record pilot.

Sources and caveats

About this article: Researched, written and edited by Umashankar Triplicane Dwarakanathan, with AI research assistance; every figure is meant to trace to the primary source cited. See the Editorial Policy for how sourcing, AI use and corrections work.