Direct Benefit Transfer in India: ₹53 Lakh Crore and What Still Leaks

In a block office in Barmer, Rajasthan, the most common complaint on a Monday morning is not that a benefit was cut. It is that the money went somewhere. A widow's pension landed in an old account she no longer uses; a farmer's instalment bounced because his Aadhaar was linked to a different bank; a ration card was struck off because the fingerprint reader could not read hands worn down by forty years of work. Direct Benefit Transfer, the machinery behind all three, is the largest government payments system in the world by number of recipients. It is also the place where India's welfare state meets the limits of its own data.

This piece looks at DBT as a payments system rather than a political slogan: how much it moves, how a rupee travels from a ministry to a bank account, what the official savings figure actually measures, and where the gaps sit in 2026.

The scale, in the government's own numbers

The DBT Bharat portal run by the DBT Mission in the Cabinet Secretariat is the official scoreboard. At the time of writing it shows cumulative transfers of ₹53,79,284 crore, a little under ₹54 lakh crore, routed through 320 schemes across 56 ministries. For the current financial year, 2026-27, it reports ₹2,61,900 crore moved in 322 crore individual transactions so far.

Two numbers explain why this matters. The first is the shift in what subsidies cost the state. A Finance Ministry assessment published by PIB in April 2025, drawing on work by the BlueKraft Digital Foundation, found subsidies fell from an average of 16% of total central expenditure in 2009-2013 to 9% in 2023-24, even as beneficiary coverage rose sixteen-fold, from 11 crore to 176 crore. The second is the size of individual schemes. PM-KISAN alone released its 23rd instalment in June 2026, more than ₹18,880 crore to over 9.44 crore farmers, taking cumulative payouts above ₹4.46 lakh crore since the scheme began in February 2019.

The 176 crore figure is beneficiary records across schemes, not people. One household can appear in PDS, LPG, PM-KISAN and a scholarship scheme at once. That distinction matters later.

How a DBT rupee actually travels

DBT is often described through the JAM trinity: Jan Dhan accounts, Aadhaar identity and mobile phones. That is the consumer-facing story. The plumbing is more specific, and most failures happen in it.

1. Sanction. A ministry or state department approves a beneficiary list and a payment file. Since a December 2014 directive, central schemes must run DBT through the Public Financial Management System (PFMS), and a 2017 Department of Expenditure order made onboarding mandatory for central sector schemes.
2. Routing. PFMS pushes the file to the sponsor bank. A May 2017 memorandum requires DBT payments to be routed through NPCI.
3. Address resolution. If the payment is Aadhaar-based, NPCI's mapper looks up which bank account is currently linked to that Aadhaar number. This is the Aadhaar Payments Bridge (APB). Whichever account was seeded most recently wins.
4. Credit. The destination bank credits the account and returns a success or a reason code. Failed credits come back to PFMS and, in theory, to the department for correction.
5. Cash-out. The citizen withdraws at a branch, an ATM, or a business correspondent using Aadhaar-enabled payments.

The DBT Mission's own payments resource page is a useful map of the rules: it lists the APB standard operating procedure, the mapper notifications from 2017, and a separate set of circulars on avoidable returns and payment failures. The existence of that last folder tells you something. Failure handling has been a standing agenda item since at least a January 2019 inter-ministerial meeting.

Step 3 is the clever part and the fragile part. Mapping by Aadhaar means a government does not need to collect and verify a bank account number for each person. It also means a single seeding change, sometimes done by a bank employee during an unrelated account opening, can silently redirect a pension.

The savings claim, read carefully

The headline efficiency number is the DBT Mission's estimated gains table: ₹5,14,201.92 crore up to March 2025, of which ₹83,063.87 crore was added in 2024-25 alone. The breakdown is revealing.

Three things stand out. Food subsidy accounts for about 61% of the total, so the national number is mostly a story about ration cards. Several schemes show zero new gains in 2024-25, which suggests the estimates are updated unevenly. And the page gives no formula. The figures are counterfactuals: what the government would have spent had the deleted records kept drawing benefits. Some of those records were ghosts and duplicates. Some were real people who failed a verification step. The table cannot tell them apart, and the government does not claim it can.

A deleted record is a saving on the ledger whether it was a ghost or a grandmother.

That is not an argument that the savings are fake. Independent research has found real effects. The World Bank's G2Px team cites a 41% fall in leakages in employment and pension programmes after biometric payments were introduced, from a 2016 study by Muralidharan and co-authors, and a 43% reduction in payment delays for a maternal health cash transfer after it went digital. The honest reading is that DBT saved a lot of money and also pushed some eligible people out, and the official figure blends both.

Worked example: one household's DBT year

Take a hypothetical smallholder family in Nashik district, Maharashtra: a farmer, his wife aged 38, and two children. Using scheme rules published by the government:

• PM-KISAN: ₹6,000 a year, paid as three instalments of ₹2,000. The arithmetic of the June 2026 release checks out: ₹18,880 crore divided across 9.44 crore farmers is ₹2,000 each.
• Maharashtra's Ladki Bahin scheme: ₹1,500 a month for eligible women aged 21 to 65 in families earning up to ₹2.5 lakh a year, or ₹18,000 a year.

That is ₹24,000 a year arriving by DBT, before LPG subsidy, ration entitlements or any scholarship. Now introduce one failure. If the wife's eKYC is not completed, her ₹1,500 stops. Over a year, the household loses ₹18,000, three times what PM-KISAN pays. If the farmer's Aadhaar was re-seeded to a new account at a cooperative bank he rarely visits, the PM-KISAN money still arrives, just somewhere he does not look. Neither failure shows up as a refusal. Both show up, in aggregate, as savings or as successful credits.

This is not hypothetical at scale. A Maharashtra verification drive reported by Business Standard in July 2026 removed over 9.2 million beneficiaries from Ladki Bahin. Roughly 6.2 million of those failed mandatory eKYC; about 1.6 million exceeded the income limit; around 29,000 were men. The same article covered a CAG finding that the state department overspent its authorised budget by ₹3,541 crore. Both stories are true at once: the scheme had genuine leakage, and its biggest single deletion category was a paperwork step.

Where it still leaks, and who falls through

Leakage in a modern DBT system rarely looks like a middleman skimming cash. It looks like this:

• Exclusion by mandate. MGNREGS made Aadhaar-based payment mandatory from 1 January 2024. Data compiled by the NREGA Sangharsh Morcha and reported by Down To Earth showed that as of 27 December 2023, 34.8% of registered workers and 12.7% of active workers were not eligible for the Aadhaar payment route. The ministry's figure was that 98.31% of active workers were seeded. The gap between "seeded" and "eligible" is where wages stall.
• Deletions in bulk. The same reporting cites job card deletion rates of 19% of workers in 2022-23. The DBT Mission's table credits 1.32 crore removed job cards between FY22 and FY25 as a saving.
• Mis-direction. Money credited to the wrong account counts as a success in the system's own data.
• Recovery. PM-KISAN has recovered ₹416.75 crore from ineligible beneficiaries as of December 2025, according to a PIB backgrounder, a small figure next to the ₹22,106 crore in estimated gains from removing 2.1 crore records.

For families abroad supporting parents in India, these failures are practical. A parent's pension or ration entitlement can quietly stop, and the fix usually requires a visit to a bank or a common service centre. Our guide on caring for parents from abroad covers the paperwork side; UIDAI's bank mapper lets anyone check which account an Aadhaar is currently linked to.

The last mile is still cash

DBT digitised the government's side of the transaction. The citizen's side often still ends with a withdrawal. Jan Dhan accounts, which passed 56 crore by the scheme's eleventh anniversary in August 2025, gave almost every household a destination for the money. Whether they then use it digitally is a different question, and the answer varies sharply by state, gender and age.

This is where India's payments stack has moved faster than its welfare stack. UPI processes over 800 million payments a day, but a pension credited to an Aadhaar-seeded account is usually withdrawn as cash through a business correspondent with a fingerprint reader. The open question for the next phase is whether benefits can be delivered in forms people spend directly, from e-RUPI vouchers to programmable e-rupee, without rebuilding the exclusion problems in a new format.

For a wider view of how India's identity, payments and data layers fit together, see our explainer on digital public infrastructure and India Stack.

What to watch next

• Next PM-KISAN instalment. The scheme released its 22nd instalment in March 2026 and its 23rd in June 2026. The next release on that cycle is the clearest monthly test of DBT capacity: around 9.4 crore credits in one day.
• The DBT Mission's update to estimated gains for 2025-26. The current table runs to March 2025. Watch whether new entries separate verified ghosts from verification failures. They currently do not.
• Union Budget 2027-28, expected on 1 February 2027. PM-KISAN was allocated ₹60,000 crore for 2026-27. Any shift in subsidy design, especially for food and fertiliser, will flow straight through DBT.
• Maharashtra's Ladki Bahin re-verification. The state cut its allocation to ₹26,500 crore for the current year from ₹36,000 crore in 2025-26. How many removed beneficiaries are reinstated after eKYC will be the best natural experiment India has on exclusion versus leakage.

DBT has made India's welfare spending far more traceable than it was in 2013. Being traceable is a different thing from being accurate. The next gains will come less from deleting records and more from making sure the right ones keep getting paid.

This article explains public schemes and payment systems. It is not legal or financial advice. Eligibility rules change; check the relevant scheme portal or a local government office for your situation.

About this researchMuffinPay InsightsIndependent analysis of stablecoins, CBDCs, cross-border payments and digital public infrastructure. Every figure is linked to its source. Read our editorial standards →