The Capex Budget and the Capacity to Spend It

Pritiraj BrahmaFounding Partner, Pritiraj & Partners · 5 February 2026

₹12.22 lakh crore of capital expenditure against a 4.3 per cent deficit. The Union missed its own capital target by ₹93,000 crore in 2024-25 and by roughly ₹33,000 crore in 2025-26 — a gap that is real, measurable and narrowing. This budget makes a coherent bet and hands the binding constraint to the states.

ProvenancePublished 5 February 2026; revised 28 August 2026 to test the absorption claim against published accounts rather than assert it. Budget aggregates are from the Union Budget documents for 2025-26 and 2026-27. Actual capital expenditure for 2023-24 and in-year spending shares are from Controller General of Accounts monthly accounts as reported; the 2025-26 comparison at 58.7 per cent against 46.2 per cent is from Observer Research Foundation analysis of those accounts, January 2026. The 2025-26 revised estimate of about ₹10.88 lakh crore is derived from the 12.3 per cent increase stated in the 2026-27 budget and is approximate. Nominal GDP assumptions are as stated in the respective budgets.

A budget can allocate capital in a day. Building the thing the capital pays for takes a district four years. The gap is not a theory: the Union missed its own capital target by ₹93,000 crore in 2024-25 and by roughly ₹33,000 crore in 2025-26. The 2026-27 budget raises the allocation by 12.3 per cent and routes more of it through states. Whether that becomes assets or arrears is a question about district machinery, not about the deficit number.

The Union Budget for 2026-27, presented on 1 February 2026, is unusually legible. It does one thing with conviction: it spends on assets while narrowing the deficit. Total expenditure is pegged at ₹53.47 lakh crore, capital expenditure at ₹12.22 lakh crore — about 12.3 per cent above the revised estimate for 2025-26 — and effective capital expenditure, which counts grants to states used for asset creation, at ₹17.15 lakh crore, roughly 4.4 per cent of GDP. The fiscal deficit is targeted at 4.3 per cent against 4.4 per cent in the revised estimate, and central government debt is projected to fall to 55.6 per cent of GDP from 56.1 per cent.

That is a coherent position and deserves assessment on its own terms. The test is not whether the allocation is large. It is whether the allocation has historically been spent.

Six findings anchor this analysis:

1. The composition shift is the substance. Total spending rises about 8.8 per cent over the 2025-26 revised estimate while capital expenditure rises 12.3 per cent. Capital spending is growing faster than the budget containing it, sustained now across several budgets — a deliberate change in what the Union government is for, and the clearest thing about this document.
2. The absorption gap is not a hypothesis; it is in the budget documents. Capital expenditure for 2024-25 was budgeted at ₹11.11 lakh crore and revised down to ₹10.18 lakh crore — a shortfall of ₹93,000 crore, against actual spending of ₹9.49 lakh crore in 2023-24. The 2025-26 allocation of ₹11.21 lakh crore was therefore only about 0.9 per cent above the previous year’s budget estimate, while reading as a 10 per cent increase over the revised one. The headline growth rate depends on which base is quoted, and the base that flatters is the one that was missed.
3. The gap is narrowing, and that matters as much as its existence. The 12.3 per cent increase stated for 2026-27 implies a 2025-26 revised estimate of about ₹10.88 lakh crore against a budget estimate of ₹11.21 lakh crore — a shortfall of roughly ₹33,000 crore, a third of the previous year’s. On execution pace the improvement is sharper: analysis of Controller General of Accounts data published in January 2026 put capital spending at 58.7 per cent of the budget estimate against 46.2 per cent at the comparable point a year earlier.
4. The 2024-25 collapse identifies the mechanism. Union capital expenditure contracted 12.3 per cent in the first eight months of 2024-25 against a budgeted increase of 17.1 per cent over the previous year’s accounts. By February 2025 cumulative spending stood at 79.7 per cent of the revised estimate against 85.0 per cent a year earlier, with that month’s capital spending down 35 per cent year on year. The proximate cause was the general election and the model code of conduct — which is to say that when the district administration is occupied with something else, capital formation stops. That is a statement about where the constraint sits.
5. The consolidation is real but slow, and rests on a growth assumption. A move from 4.4 to 4.3 per cent is a tenth of a point. The deficit has come a long way from the pandemic peak, and the remaining distance to a 4 per cent anchor is being covered on an assumption of roughly 10 per cent nominal growth. Fiscal ratios improve when the denominator cooperates; the 2025-26 budget rested on a nominal GDP estimate of ₹356.98 lakh crore, about 10.1 per cent above the preceding revised estimate. Consolidation delivered by the denominator is not the same as consolidation delivered by the numerator.
6. The risk has been relocated outside the budget. Effective capital expenditure of ₹17.15 lakh crore includes grants to states for asset creation, against ₹15.48 lakh crore budgeted the previous year. The Sixteenth Finance Commission’s report was tabled the same day, and its award discontinues revenue deficit grants while routing much of the new grant money to local bodies against performance conditions. More of India’s capital programme now depends on state and local execution, at the moment when state fiscal cushions are thinner.

The number that decides this budget

An allocation is not an asset. Between the two sit land acquisition, environmental clearance, tendering, contractor capacity, supervision and payment. Every one of those steps happens in a state, most in a district, and none is funded by the line item that pays for the structure. The figure above is what that costs. Two consecutive years in which the Union could not spend what it had allocated is not an accounting curiosity; it is ₹1.26 lakh crore of intended asset formation that did not happen on schedule.

It is also, importantly, a gap that is closing. A serious reading has to hold both facts. The 2024-25 shortfall was large and election-driven; the 2025-26 shortfall was a third of the size and the pace of spending improved markedly. Someone in the system has been working on this, and the improvement is measurable. What has not changed is that the improvement is visible only to those who go looking for it in monthly accounts, because no department publishes value sanctioned against value commissioned as a matter of routine.

What the pace tells you

The two panels above make a point that annual totals hide. Capital spending in India is heavily back-loaded, and a back-loaded programme is a fragile one: it depends on the final quarter, when contractors are paid against work certified in a rush and the incentive is to certify. A programme that reached 58.7 per cent by the same point in 2025-26 is structurally safer than one that reached 46.2 per cent, whatever the two years ended at.

The honest counter-argument to the district-capacity thesis is that this is precisely what the Infrastructure Risk Guarantee Fund is for: bringing in private developers who carry their own execution capability, alongside recycling of public-sector real estate through dedicated trusts. That is a genuine answer for large projects with bankable revenue. It is not an answer for the district road, the school building or the water scheme, which is where a large share of effective capital expenditure lands and where no private developer is bidding.

The counter-case, honestly stated

Three objections deserve a hearing.

First, routing capital through grants to states relocates the hard part, and relocation may be the correct design. District roads, school buildings and water schemes are built better by states and districts than by any Union agency, and a centre that took execution in-house to control it would deliver worse assets more slowly. Conditional grants are how federations build local infrastructure. The real objection is that state execution capacity is inadequate — a claim about states, easily mistaken for a criticism of a Union accounting convention.

Second, the shortfall may be evidence of discipline rather than incapacity. A government that revises capital expenditure down when projects are not ready, rather than releasing money against incomplete work to protect a headline, is behaving correctly. The 2024-25 revision could be read that way. The counter to the counter is the pace data: a programme that spends 46.2 per cent of its allocation by the same point and then attempts the rest is not exercising judgment, it is running late.

Third, the full income-tax rebate for individuals earning up to ₹12.75 lakh cuts against this essay’s own emphasis. It reaches households already inside the formal tax net, while most Indian income is informal — and that is also the strongest argument for it. It is the one large measure in the budget that requires no district execution whatsoever: no tendering, no supervision, no commissioning gap. In a budget whose principal risk is delivery machinery, a transfer that bypasses the machinery entirely is not a concentration problem to note in passing. It is the only instrument here with a guaranteed transmission mechanism.

What we would do

For a state government reading this budget as an opportunity rather than an announcement, five things follow.

1. Publish value sanctioned against value commissioned, per department, monthly. With the median interval between the two. This is the number that settles the argument above in either direction, and almost no state publishes it. The Union’s own gap became visible only because the Controller General of Accounts publishes monthly; states that publish nothing comparable are asking to be judged on allocation, and they will be.
2. Build the pipeline before the money is available, not after. The states that draw the largest share of central capital transfers are consistently those with cleared, tendered, shovel-ready projects on the shelf. The 2024-25 shortfall shows what happens when readiness is the binding constraint at national scale. Preparation is the competitive advantage, and it is cheap.
3. Front-load deliberately, and report quarterly against a front-loaded profile. A programme that plans to spend evenly and actually spends in the fourth quarter is certifying work under time pressure. Publish the intended quarterly profile at the start of the year and the achieved profile against it.
4. Staff the project-management function explicitly. A capital programme growing at double digits with a flat project-management cadre converts the increase into delay rather than assets. This is the cheapest line in any capital budget and the first one cut.
5. Plan the revenue side for a world without the old cushions. With revenue deficit grants discontinued and much of the new grant money conditional on performance, states that assume the previous pattern of discretionary top-ups are planning against a budget that no longer exists.

This is a competent budget with a clear theory: build assets, hold the deficit, attract technology capital. The theory is sound and the execution record is improving. Its remaining risk is that the arithmetic is done in Delhi and the work is done in a district office that nobody has funded to do it — and that when the gap opens again, as it did in an election year, no state will be publishing the number that would show it.