Announced, unspent, announced again
India budgeted ₹10.49 lakh crore for centrally sponsored schemes and other transfers in 2025-26. By the revised estimates it was ₹8.46 lakh crore. Four flagship schemes lost 70–85% of their money in-year — and then received almost exactly the same allocation again.
Jal Jeevan Mission was budgeted ₹67,000 crore in 2025-26 and revised to ₹17,000 crore. In 2026-27 it is budgeted ₹67,670 crore. Bharatnet: ₹22,000 crore, revised to ₹5,500 crore, budgeted again at ₹20,000 crore. The Research, Development and Innovation scheme: ₹20,000 crore, revised to ₹3,000 crore, budgeted again at ₹20,000 crore.
The pattern is not a cut. It is the same figure being put forward twice.
A fifth of the scheme budget was revised away
Every year the Union Budget publishes a revised estimate alongside the new one: the government’s own updated view of what the current year will actually cost. For centrally sponsored schemes and other transfers in 2025-26, that revision was downward by ₹2,03,803 crore — 19.4% of what had been budgeted.
This is not an audit finding or an outside estimate. It is the government revising its own number, in the same document that sets next year’s.
View data table
| Line | ₹ crore |
|---|---|
| Budget Estimate 2025-26 | 10,49,343 |
| Retained in Revised Estimate | 8,45,540 |
| Revised away | 2,03,803 |
The shortfall is concentrated, not spread
Most large schemes spent close to what they were given. The National Health Mission came within 0.3% of its allocation; PM-KISAN and the Modified Interest Subvention Scheme were unchanged; MGNREGS was revised slightly up. The gap sits in a handful of programmes, and in those it is enormous.
View data table
| Scheme | BE 2025-26 | RE 2025-26 | Change |
|---|---|---|---|
| Jal Jeevan Mission | 67,000 | 17,000 | −74.6% |
| PM Awas Yojana – Rural | 54,832 | 32,500 | −40.7% |
| Samagra Shiksha | 41,250 | 38,000 | −7.9% |
| National Health Mission | 37,227 | 37,100 | −0.3% |
| PM Awas Yojana – Urban | 25,794 | 7,900 | −69.4% |
| Bharatnet | 22,000 | 5,500 | −75.0% |
| Research, Development & Innovation | 20,000 | 3,000 | −85.0% |
| PM Surya Ghar Muft Bijli | 20,000 | 17,000 | −15.0% |
| Saksham Anganwadi & POSHAN 2.0 | 21,960 | 20,949 | −4.6% |
| PM-KISAN | 63,500 | 63,500 | 0% |
| MGNREGS | 86,000 | 88,000 | +2.3% |
Next year’s number is last year’s number
The striking thing is not the shortfall on its own. It is what the following budget does with it. For the four worst-hit schemes, the 2026-27 allocation is close to the 2025-26 allocation that was never spent — not to the amount actually spent.
Read down the middle column and then the right-hand one: the money reappears. PRS makes the same observation about the housing schemes, noting that “the budget allocation in 2026-27 is similar to amounts originally budgeted in 2025-26”.
View data table
| Scheme | BE 2025-26 | RE 2025-26 | BE 2026-27 |
|---|---|---|---|
| Jal Jeevan Mission | 67,000 | 17,000 | 67,670 |
| Bharatnet | 22,000 | 5,500 | 20,000 |
| Research, Development & Innovation | 20,000 | 3,000 | 20,000 |
| PM Awas Yojana – Urban | 25,794 | 7,900 | 22,025 |
MGNREGS falls 66%, and that is not the story
The scheme table shows MGNREGS falling from ₹88,000 crore in the 2025-26 revised estimates to ₹30,000 crore in 2026-27 — a 66% drop, and by far the largest fall in the table. Quoted alone it would be the headline of this dive.
It should not be. In December 2025 MGNREGA was replaced by the VB-G RAM G Act, and the new programme is allocated ₹95,692 crore in 2026-27 — the single largest scheme allocation in the Budget. Taken together the two lines are ₹1,25,692 crore against ₹88,000 crore. Whether the successor programme delivers what its predecessor did is a real question, and not one a budget line can answer. What the number plainly is not is a 66% cut to rural employment spending.
A revised estimate is still an estimate. The 2025-26 RE is the government’s mid-year view, not audited spending. Actuals will differ, and can land either side. The one column of genuine actuals here is 2024-25.
Underspending is not automatically waste, or malice. Money goes unspent because approvals are slow, because states have not sent matching shares, because tenders fail, because a scheme is genuinely ahead of demand, or because the original number was never achievable. These have very different implications and the budget line looks identical for all of them. This dive shows the gap; it does not diagnose it.
“Announced twice” is a description, not an accusation. Re-budgeting an unspent allocation is ordinary practice when a programme is delayed rather than abandoned — the work still has to be paid for. The observation worth making is narrower: a headline allocation figure tells you what a government intends to spend, and in these four cases the previous year’s identical intention was not met.
Comparisons across the MGNREGA/VB-G RAM G change are unsafe. Section 04 exists because the largest number in the table is the most misleading one. Any year-on-year rural employment comparison from 2026-27 onwards has to handle the replacement explicitly.
What this cannot tell you. Nothing here reaches the district, the worksite or the household. An unspent central allocation does not map cleanly onto services not delivered — states fund and deliver much of this themselves, and the timing of central releases and state spending do not coincide.
- Union Budget 2026-27 Analysis, PRS Legislative Research, 1 February 2026 (corrected 2 February 2026). Scheme-wise allocations, the centrally sponsored schemes aggregate, and the note on the MGNREGA replacement.
- Union Budget 2026-27, Ministry of Finance — Expenditure Profile, the underlying source for the PRS tables.
Percentage changes are ImpactMojo’s arithmetic on the published rupee figures and appear in the data tables.
ImpactMojo Data (2026). “Announced, unspent, announced again.” ImpactMojo Data Dives. Retrieved from https://impactmojo.in/DataDives/announced-unspent-announced-again.html
Have a dataset worth digging into?
Data Dives are independent investigations built from public data. If you know a number that deserves a closer look, tell us.
Pitch a Data Dive →