Government Payroll and Wage Payments in India: NREGA to VB-G RAM G

In Balangir district of western Odisha, some rural job-scheme workers told surveyors in 2025 that they had waited more than five months for wages the law said should reach them in fifteen days. The money itself moved in seconds once it moved. Everything before that moment was the problem, and that is the real story of how the Indian state pays the people who work for it.

Government payroll in India runs on two very different tracks. One is the salaried track: roughly 50 lakh central government employees and 69 lakh central pensioners, plus several crore state employees, paid monthly through treasury systems that have been digitising for two decades. The other is the daily-wage track, dominated for twenty years by the rural employment guarantee, which paid crores of casual workers through a chain of muster rolls, wage lists and fund transfer orders. On 1 July 2026 that second track changed its legal basis entirely. This piece looks at what the data says about how well both tracks work, and where the next failures are likely to sit.

The rails are almost fully digital already

The headline numbers on digitisation are not in dispute. According to a Ministry of Rural Development overview published by PIB in August 2025, the share of rural job-scheme wages paid through the electronic fund management system rose from 37% in 2013-14 to 99.94% in 2024-25. Aadhaar numbers seeded in the scheme database went from about 76 lakh in January 2014 to 13.45 crore, and 13.05 crore workers were linked to the Aadhaar Payment Bridge System, the NPCI mapper that routes a credit to whichever bank account a person's Aadhaar number currently points at.

That scale is enormous. The same document puts person-days generated in 2024-25 at 290.60 crore across 15.99 crore registered households, with women supplying 58.15% of the work, up from 48% in 2013-14. The budget allocation for 2025-26 was Rs 86,000 crore, against Rs 33,000 crore in 2013-14.

So when people talk about "digitising government wages" in India, they are describing something that has largely happened. The more useful question is what the digital rails did and did not fix.

Where the delay actually lives

A wage payment under the old scheme passed through roughly five hand-offs. A muster roll recorded attendance at the worksite. A wage list was generated after the roll closed. A fund transfer order (FTO) was created and digitally signed by two officials at the block or panchayat level. The FTO then went to the Centre, which released money through PFMS, and finally the credit landed in the worker's account through the Aadhaar bridge or an account-based transfer.

Official reporting measures the first half of that chain. In a Lok Sabha reply on 22 July 2025, the ministry said over 96% of FTOs had been generated within 15 days of muster roll closure in each of the previous five years, and that 100% of wage payments went directly to beneficiaries through DBT. The PIB overview put on-time FTO generation at 97.81% by March 2025.

Those figures are probably accurate, and they are also the wrong yardstick. An FTO generated on day 12 says nothing about when the Centre actually pushed the money, or whether the credit bounced because a bank account was no longer correctly mapped. Researchers at LibTech India have argued for years that the second stage, between FTO signing and credit, is where most delay accumulates, and that it was historically excluded from delay compensation calculations. Their 2024-25 national tracker collected worker testimony of waits of up to three months in Karnataka and Uttar Pradesh and over five months in parts of Odisha.

The money that never arrived is measurable too. Unpaid wages under the scheme rose from Rs 512.75 crore in 2020-21 to Rs 974.38 crore in 2024-25, according to ministry data compiled by IndiaSpend as of March 2025. The ministry told Parliament in July 2025 that all pending wage liabilities up to 2024-25 had since been cleared. Both statements can be true: liabilities get cleared in lumps, after the household has already borrowed to cover the gap.

A payment system is only as fast as its slowest signature, and in government payroll the slowest signature is usually a fund release, not a bank.

ABPS: the identity layer that cut both ways

The Aadhaar-based payment system has existed in the scheme since 2017, but the government made it mandatory for wage payments from 1 January 2024, after several extensions from an original February 2023 date. The logic is sound on paper. A worker's Aadhaar becomes their financial address, so a changed or closed bank account no longer breaks the payment as long as the NPCI mapper is updated. It also makes duplicate or fictitious beneficiaries harder to sustain.

The cost was eligibility. A worker whose Aadhaar was not seeded, whose name was spelled differently on the job card, or whose bank had not completed mapping could not be paid through ABPS at all. Coverage reported by Rau's IAS at the time noted that 12.7% of active workers were still ineligible in December 2023 and cited about 7.6 crore job-card deletions since April 2022, often linked to such mismatches. By April 2025, LibTech's tracker put ABPS ineligibility at 27.5% of all registered workers but only 1.5% of active workers.

That gap between "all" and "active" is the whole debate in one number. The government reads it as a clean-up success: the people actually working are almost all payable. Critics read the same figure as evidence that people who could not get linked simply stopped being counted as workers. The trend in participation is consistent with both readings. Workers who availed employment fell from 11.17 crore in 2020-21 to 7.88 crore in 2024-25, per the Lok Sabha reply, though the 2020-21 peak was inflated by pandemic-era demand.

The new law: 125 days, shared bills, a weekly clock

Parliament passed the Viksit Bharat Guarantee for Rozgar and Ajeevika Mission (Gramin) Act in December 2025, and it replaced the 2005 Act from 1 July 2026. According to a PIB explainer from May 2026, the guarantee rises to 125 days per household per year, with states allowed to notify a pause of up to 60 days during peak sowing and harvest. Wages are to be paid weekly, or within 15 days of muster roll closure, with compensation of 0.05% of unpaid wages per day of delay beyond the 16th day. The central share for 2026-27 is Rs 95,692.31 crore, and the total outlay is expected to exceed Rs 1.51 lakh crore once state shares are counted. The administrative expenditure ceiling goes from 6% to 9%.

The change that matters most for payments is the funding split. A Drishti summary of the Act describes a 60:40 Centre-state ratio for most states and 90:10 for northeastern and Himalayan states, where the old scheme had the Centre carrying the entire unskilled wage bill. In practice, every wage payment now needs two treasuries to have cash ready at the same time.

Worked example: what a late wage is worth

Take a worker on a notified wage of Rs 300 a day who completes 12 days on a fortnightly muster roll. The wage due is Rs 3,600. Suppose the money arrives 46 days after the muster closes, roughly the kind of wait workers described to LibTech in Odisha and Karnataka, and not an extreme one.

• Days of delay counted for compensation: 46 minus 16 = 30 days.
• Compensation rate: 0.05% per day x 30 days = 1.5%.
• Compensation owed: 1.5% of Rs 3,600 = Rs 54.

Rs 54 for a month-long delay works out to an annualised rate of about 18%. That sounds respectable until you compare it with what a rural household actually pays to bridge a month without wages from an informal lender, which is frequently far higher. And it assumes the compensation is calculated, approved and credited, which has historically been the weakest link. Our view: the rate is too low to change official behaviour on its own. What will change behaviour is whether the weekly clock gets measured end to end, from muster closure to credit, and published by state.

Salaried payroll: quieter, but not finished

On the salaried side, digitisation looks less dramatic because it happened earlier, but the numbers are large. Karnataka's HRMS, integrated with its Khajane-II treasury, covers about 5.03 lakh government employees and generates around 7.9 lakh pay bills a year, according to a Karnataka presentation to a CAG conference of state finance secretaries in September 2025. From 1 June 2025 the state stopped sending printed copies of roughly 7.5 lakh salary-bill vouchers a year to the Accountant General. Telangana is still entering records for about 3.48 lakh employees into its e-HRMS, with a 15 October 2026 target set in a review this month. Tamil Nadu's IFHRMS, sanctioned at Rs 288.91 crore in 2018, was designed for about 29,000 drawing and disbursing officers and 9 lakh digitised service registers.

Pensions show what end-to-end digitisation can do. The defence ministry's SPARSH system, launched in October 2020, handles both sanction and disbursement on one platform and had onboarded about 31 lakh of 32 lakh defence pensioners by April 2025, according to reporting on defence ministry figures. It replaced an arrangement involving three agencies and over 45,000 public interfaces. That is the model wage schemes have not reached: one ledger owning the entire path from entitlement to credit.

The fiscal pressure behind all of this is rising. The CAG's decade review found states' committed spending on salaries, pensions and interest grew 2.49 times between 2013-14 and 2022-23, to about Rs 15.64 lakh crore, with salaries the largest item in 19 states, as Business Standard reported. When the payroll is that large a share of the budget, a treasury that cannot see its liabilities in real time ends up delaying whatever is easiest to delay. Historically, that has been wages for people with the least power to complain.

Why this matters beyond India

Many governments are now trying to replicate India's model of identity-linked payments; we cover the broader pattern in G2P payments around the world. The Indian lesson is uncomfortable but useful. Identity-based routing removes one class of failure, the stale bank account, and introduces another, the identity mismatch that blocks a payment entirely. It also tends to make delays look like someone else's problem, because each stage reports its own on-time rate. Gulf states solved a different version of this with mandated wage-file monitoring, described in our explainer on wage protection systems and payroll cards, which put the employer's payment date itself under surveillance.

For readers following India's welfare plumbing more broadly, our guides to direct benefit transfer and India's digital public infrastructure cover the shared rails, and how state treasuries move money explains why the new 60:40 split makes state cash management the critical variable.

What to watch next

• 15 October 2026: Telangana's deadline for completing e-HRMS records for its roughly 3.48 lakh employees. Slippage would be an early signal of how hard state payroll consolidation still is.
• Q3 and Q4 of 2026-27: the first full quarters of VB-G RAM G payments. Watch whether states publish end-to-end payment times, not just FTO generation rates, and whether any state's 40% share arrives late enough to trigger compensation at scale.
• Around April 2027: the 8th Central Pay Commission, whose terms of reference the Cabinet approved in late October 2025 with an 18-month deadline, is due to report. Its recommendations, expected to take effect from 1 January 2026 per The Tribune, will mean arrears calculations for about 50 lakh employees and 69 lakh pensioners, a large stress test for payroll systems.
• 1 February 2027: the Union Budget for 2027-28 will show whether the central allocation for rural employment holds near Rs 95,692 crore once actual state co-funding behaviour is known.

Figures are from government releases and independent trackers as cited; scheme data is revised frequently, so check the latest ministry dashboards for current values.

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 →