The Discom Turnaround — and What It Does Not Yet Prove

Pritiraj BrahmaFounding Partner, Pritiraj & Partners · 2 April 2026

India’s power distributors posted a collective profit of ₹2,701 crore in FY25 — the first in over a decade — and closed the gap between what a unit costs and what it earns from ₹0.69 to ₹0.06. A real turnaround, achieved by conditioning money on verified administrative acts. Whether the design or the cycle did the work is a question the sector’s own data could answer and does not.

ProvenancePublished 2 April 2026; revised 28 August 2026 to separate the operating and financing questions, add a sixth finding and two figures, and remove a framing the essay’s own counter-case had refuted. Loss, cost-recovery and profit figures are from power sector utilities data for the years shown; a parliamentary reply citing 16.16 per cent for FY25 against 15.04 per cent is noted rather than reconciled. Scheme sanction and smart-meter figures are as published, with meter installations as at March 2026. Installed capacity is as at January 2026. The Sixteenth Finance Commission reference is to its award and recommendations on deficit and debt definitions.

For the first time in over a decade India’s electricity distributors made money — a collective ₹2,701 crore in FY25, against losses exceeding ₹67,000 crore in 2013-14. The measure that matters more: the gap between what a unit costs and what it earns has fallen from ₹0.69 to ₹0.06. The instrument was conditionality, not grant, and that is the most transferable lesson in Indian governance this decade. It is also a first profitable year, not a solved problem.

India’s distribution companies have been the weakest link in its power system for as long as the system has existed. They buy electricity, wheel it, bill consumers, and — across most of the country, for most of the last two decades — fail to recover what they spend. In FY25 that changed at the aggregate level.

This is a genuine achievement and it should be said plainly before it is qualified.

Six findings anchor this analysis:

1. Losses fell to their lowest recorded level. Aggregate technical and commercial losses — the measure that combines theft, faulty metering, unbilled supply and failed collections — fell to 15.04 per cent in FY25, from 15.97 per cent the previous year, 21.91 per cent in FY21, 22.62 per cent in 2013-14 and 27.34 per cent in 2008-09. A parliamentary reply cited 16.16 per cent for FY25 against the utilities figure of 15.04 per cent; both are stated rather than reconciled.
2. Cost recovery is now nearly complete on average, and this is the finding that matters. The national gap between average cost of supply and average revenue realised has narrowed to ₹0.06 per unit from ₹0.69. On the aggregate, Indian discoms very nearly recover what electricity costs them — which is the condition that determines whether the sector adds to its deficit or stops.
3. The instrument was conditionality, not grant. The Revamped Distribution Sector Scheme ties fund release to measurable performance — feeder metering, loss reduction, timely payment of government dues, regular tariff orders, no creation of new regulatory assets. Projects worth about ₹2.83 lakh crore have been sanctioned under it, including ₹1.53 lakh crore for distribution infrastructure, and 59.7 million smart meters had been installed by March 2026.
4. Payment discipline did work the metering could not. Rules tightening late-payment surcharges allowed legacy dues to be cleared in instalments while preventing new arrears from snowballing, which is why collection efficiency improved sharply from 2022. Metering fixed what could be measured; the payment rules fixed what could be enforced. Attributing the turnaround to either alone misreads it.
5. The accumulated deficit is a financing decision that nobody has taken. Accumulated losses stand at about ₹6.47 lakh crore. This is the balance-sheet residue of past under-recovery, not a measure of current performance, and it will not be resolved by operating profit at this scale. Whether it is restructured, absorbed or amortised, no published plan addresses it.
6. The national average conceals the distribution, and nothing requires it to be disclosed. Aggregate figures of this kind are dominated by the best-performing utilities, and individual discoms remain far from the scheme’s own band of 12 to 15 per cent. A handful of large, poorly performing utilities can absorb the sector’s entire aggregate profit. There is no single national table publishing losses, cost-recovery gap, days payable and metering coverage per utility, which means the average is doing accountability work it cannot do.

The loss curve

India has attempted discom reform repeatedly, most recently through UDAY, and the pattern has been consistent: state governments take the debt onto their books, the utility gets a clean balance sheet, tariffs remain politically frozen, losses re-accumulate, and a new scheme is announced a few years later.

What is different this time is that the money was made contingent on things that can be verified — a meter installed, a due paid, a tariff order issued on time. That is a design choice worth generalising well beyond electricity: a transfer conditioned on an audited administrative act is far harder to convert into a bailout than a transfer conditioned on a promise.

Flow and stock are different problems

Keeping these two quantities apart is the analytical discipline this subject usually lacks. The ₹6.47 lakh crore is real and somebody will pay it. But it says nothing about whether a utility is well run today, and a sector that has closed a ₹0.63 per-unit recovery gap has done the hard operating work. The failure now is that no plan for the stock has been published, not that the stock exists.

What the turnaround has not yet shown

Three things remain unproven, and a serious assessment should hold them open.

Whether it survives a tariff cycle. Near-complete cost recovery has been achieved in a particular fuel-cost and demand environment. The test of a discom’s finances is not a good year; it is whether tariffs move when costs move. That is a political act, and it has not been tested under stress.

Whether the laggards converge. The improvement is national and the accountability is not. Without utility-level publication and consequence, the average will keep improving while the outliers stay stuck.

Whether the system can carry what is coming. Installed generation capacity reached 520.51 GW by January 2026, with 296.388 GW added since April 2014, and the Draft National Electricity Policy 2026 signals a pivot toward market-based procurement and away from rigid long-term coal power purchase agreements. A utility that has just learned to recover its costs under long-term contracts now has to learn to buy in a market — a materially harder commercial skill.

The counter-case, honestly stated

Three arguments cut against this analysis, and the second is the one that constrains what can be concluded.

First, on the framing this essay has deliberately avoided: dividing ₹6.47 lakh crore of accumulated losses by a single year’s ₹2,701 crore profit yields roughly two hundred and forty years, and the figure is arresting and analytically empty. It assumes profit fixed in perpetuity, a proposition nobody advances. We name it because it circulates, and because an essay that used it would be making the stock-versus-flow error its own second finding warns against.

Second, the causal weight placed on conditionality may be too heavy. Losses fell from 21.91 per cent in FY21 to 15.04 per cent in FY25 — a window that coincides with the post-pandemic recovery in demand and collections, a particular fuel-cost environment, and the separate tightening of late-payment surcharge rules. Distinguishing what the scheme caused from what the cycle delivered requires a comparison this essay does not attempt: performance in states that moved early against those that moved late. That matters more here than elsewhere, because the transferable lesson being drawn — condition money on verified administrative acts — is precisely the claim that needs the cycle ruled out before it is generalised across Indian governance.

Third, our own central recommendation relocates the deficit rather than removing it. Paying subsidy as a visible, on-time budget line instead of a suppressed tariff is correct accounting, and it converts a utility’s loss into a state’s expenditure. The Sixteenth Finance Commission has fixed the state fiscal deficit ceiling at 3 per cent of gross state domestic product and called for the definitions of deficit and debt to be widened to capture what previously sat outside them. A state that makes its power subsidy explicit and punctual is moving an obligation from a parastatal balance sheet onto the one now being fenced. The reform is right. It is not free, and the fiscal regime it collides with is the one this Review has itself identified as binding.

What we would do

1. Publish utility-level performance in a single national table, quarterly. Losses, cost-recovery gap, days payable and smart-meter coverage, per discom. The national average is a poor management instrument; it tells a good utility nothing and lets a bad one hide. This is also the comparison that would let anyone test the causal claim in the second objection above.
2. Put government departments on prepaid meters first. Public offices are among the most persistent defaulters, the fix is administrative rather than political, and it removes the discom’s standard excuse.
3. Make the tariff order automatic and the subsidy explicit. Where a state wishes to subsidise a category, it should do so as a visible budget line paid on time, not as a suppressed tariff absorbed by the utility. This is the single reform that would make the FY25 result durable, and it should be adopted with the fiscal consequence stated rather than concealed.
4. Build commercial capability before the market pivot lands. If procurement is to move to market-based mechanisms, discoms need trading and forecasting capacity they do not currently have. Buying badly in a market is more expensive than buying rigidly under a contract.
5. State the plan for the ₹6.47 lakh crore. Whether it is restructured, absorbed or amortised, it should be addressed openly rather than carried indefinitely as a number nobody plans against. The absence of a plan, not the size of the number, is the finding.

A sector that lost more than ₹67,000 crore in a year now makes a small profit and loses one unit in seven instead of one in four. That is real progress, achieved by conditioning money on verified administrative acts. Whether the design or the cycle did the work is a question the sector’s own data could answer and does not, because nobody is required to publish it utility by utility.