Sharing Distribution Networks, but Not the Costs 

Sharing Distribution Networks, but Not the Costs 

India's new distribution policy solves one problem — and quietly creates another

The draft National Electricity Policy (NEP) 2026 proposes something India's power sector has circled for two decades without landing: shared distribution networks. The idea is to let multiple suppliers compete for the same consumer over the same wires, instead of forcing every new entrant to build a parallel network. A Distribution System Operator (DSO) is proposed to make this sharing work, and to integrate rooftop solar, storage and Vehicle-to-Grid resources along the way.

On its face, this fixes a real design flaw. But look closer, and it opens a harder question that the policy has not yet answered: if one company builds, owns and maintains the network — and a rival is simply allowed to use it to take that company's best customers — who pays for what's left behind?

That question is not academic. India's distribution sector is already financially stressed, with accumulated losses, ageing infrastructure and thin regulatory margins in many states. A reform that gets the cost allocation wrong won't just be unfair — it could tip already-weak DISCOMs further into distress, at exactly the moment the sector needs capital for the energy transition.

Why parallel wires never worked

The Electricity Act, 2003 already allows more than one distribution licensee in the same area. But it assumed each licensee would build its own network. Two decades on, that model has produced almost no competition — Mumbai is the lone real example — because building a second set of poles, feeders and substations is capital-intensive and, in most cities, simply irrational.

The government has now acknowledged this. The proposed Electricity (Amendment) Bill, 2025 would let one licensee use another's network on payment of SERC-determined charges. The Draft NEP 2026 takes the same direction, explicitly framing shared networks as a way to boost competition and efficiency while avoiding duplicated infrastructure.

Conceptually, this shifts the industry from competition between networks to competition over a common network. That is the right shift to make.

The economics genuinely work — up to a point

Distribution networks are natural monopolies. It makes no more sense to run three sets of poles down one street than it does for three telecom operators to dig three separate trenches to your door. Letting the network stay a regulated monopoly while supply becomes competitive is a coherent model, and the draft NEP deserves credit for recognising it.

The economics are analogous to other network industries: competing telecom operators don't each dig a separate trench to your door, yet you can still choose your carrier. If India can pull off the electricity equivalent — one set of wires, many competing suppliers — it would be a genuine structural advance over two decades of stalled distribution reform.

But avoiding duplicate wires is the easy half of the reform. The harder half is what happens next.

Who actually paid for this network?

Distribution networks are not free. Every pole, transformer, substation and meter has been financed over decades and recovered through regulated tariffs paid by existing consumers. The incumbent DISCOM has carried the obligation to maintain that network, augment it, and serve everyone — including consumers nobody wants.

A new entrant doesn't have to build any of this. It simply pays a regulated charge to use it. Fair enough — if that charge reflects the network's true economic cost. But that "if" is doing a lot of work, because it immediately raises questions the policy has not settled:

●        Who bears the unrecovered historical cost of the network?

●        Who pays for future augmentation?

●        Who absorbs stranded assets when profitable consumers leave?

●        Who carries the fixed cost of existing power purchase agreements (PPAs) tied to those consumers?

●        Who retains the universal service obligation?

None of these can be waved away with a generic principle of "non-discriminatory access."

The cherry-picking problem

A new supplier will chase the consumers who are cheapest to serve and most profitable to keep — typically large industrial and commercial loads. The incumbent DISCOM has never had that luxury: it has served everyone, including domestic, agricultural and rural consumers who may cost more to serve than they pay.

The government insists universal service obligations will remain, and that cherry-picking will be prevented. That's necessary, but it doesn't answer the deeper question: are the economic consequences of a consumer switching supplier actually symmetrical?

When a new supplier wins a large industrial load, the incumbent loses the revenue — but the feeder, the transformer, the substation, the maintenance obligation and often the underlying PPA commitment all stay exactly where they were. Revenue follows the consumer. A large share of the cost does not.

That mismatch is arguably the single most important design problem in shared distribution.

Not "taxation without representation" — but close

There's a tempting historical parallel here, and it needs handling carefully. "Taxation without representation" was the colonial grievance that taxes were imposed without a vote in the body imposing them. A network charge is not a tax, and today's DISCOM consumers are not colonists.

But the underlying asymmetry rhymes. A network built and paid for by one set of consumers, over decades, can end up being used by a rival supplier to win the same consumers' most valuable neighbours — while the original consumers keep carrying the network's unrecovered cost. Put plainly:

The consumers who financed the network bear the cost, while another supplier captures the commercial benefit of using it.

A more precise framing than the colonial analogy: network access without corresponding responsibility.

Competition should not mean socialising the incumbent's costs

Picture a DISCOM serving a million consumers on a large regulated asset base. A competing supplier enters and wins the 200,000 most attractive ones. The physical network barely changes size. If the network-use charge is calculated only on incremental operating cost, the incumbent is left recovering a largely fixed network cost from a shrunken, less profitable consumer base.

That sets off a familiar spiral: consumer migration → falling incumbent sales → rising average network cost → higher tariffs → more migration. It's a textbook stranded-cost problem, and it means competition could end up weakening — not strengthening — the sector's financial health, unless the policy explicitly separates network costs, supply costs, legacy liabilities and universal-service costs into distinct, transparently recovered buckets.

What should the network charge actually recover?

This is where the regulatory detail matters more than the headline principle. A network-use charge cannot simply be an administratively set wheeling fee. At minimum, the methodology needs to address the remaining regulated value of existing assets and how it's recovered; depreciation and return on capital for the network owner; O&M allocation; technical and commercial losses; who funds incremental augmentation when a new supplier brings in additional load; the redundancy needed to hold reliability standards; metering, billing and settlement infrastructure; DSO operating costs; and — critically — stranded costs when assets or contracts become underused after consumers migrate.

These aren't bookkeeping footnotes. They decide whether the competitive model is financially sustainable at all. Get the network charge wrong in either direction, and the reform fails: price it too high, and no supplier enters — competition never materialises. Price it too low, relative to the network's true cost, and the incumbent's balance sheet absorbs the difference, one migrated consumer at a time.

The PPA problem is even harder than the network problem

The incumbent's other major fixed-cost exposure is its portfolio of long-term power purchase agreements. A PPA's fixed or capacity charge typically doesn't shrink just because a consumer switches supplier — the incumbent can remain liable for it even after the load, and the revenue that came with it, has gone.

PRS Legislative Research has flagged exactly this in its review of the Draft Electricity (Amendment) Bill, 2025: incumbent DISCOMs may stay on the hook for fixed PPA charges after consumers migrate, with no clear mechanism yet for reallocating that liability. This means distribution competition cannot be designed as a network-access reform alone. It is, just as much, a power-procurement reform — and the current draft doesn't fully grapple with that.

The missing piece: a genuinely neutral DSO

The logical endpoint of shared networks isn't "several DISCOMs using each other's wires." It's a cleaner architecture: competitive supply, over a regulated common network, run by an independent system operator.

This is where the proposed DSO matters. The Draft NEP 2026 envisages it facilitating network sharing, integrating distributed energy resources, and enabling local trading — a real foundation for a distribution market, if built correctly.

But it only works if the DSO is functionally independent of every competing supplier. If the incumbent DISCOM is simultaneously the network owner and a competing retailer, the conflict of interest is structural, not hypothetical: it controls the very infrastructure its rivals depend on. The DSO's responsibilities — connection management, congestion and loss management, settlement data, DER integration, non-discriminatory access — only mean something if it has no commercial stake in who wins the consumer.

This is a different, more demanding architecture than simple "carrier-content separation," where one regulated entity owns the wires and multiple suppliers compete over them. NEP 2026 gestures toward this model without yet committing to it. That gap is worth closing, because as long as the network owner is also a supplier, network neutrality and supply competition will keep pulling in opposite directions.

Who owns the consumer — and who owns the subsidy?

Under the classic vertically integrated DISCOM, the consumer is effectively an asset of the utility: it procures power, meters, bills and maintains the connection. Competitive supply breaks this apart — the consumer becomes a customer of a supplier, while the physical connection belongs to the network. That's the real conceptual shift NEP 2026 is proposing, whether or not it says so explicitly.

It gets more complicated where cross-subsidy is significant. If industrial consumers migrate disproportionately, the incumbent's ability to recover cost from its remaining portfolio changes materially. The policy needs to cleanly separate cross-subsidy attached to the consumer from network cost attached to the infrastructure — and ideally, any subsidy a state wants to guarantee should be explicitly and separately funded by the state, not buried inside network or supply accounts. Left unseparated, consumer migration could destabilise the entire tariff structure.

Don't punish the incumbent for having built the thing

There's a fairness dimension too easily lost in competition debates. The incumbent DISCOM didn't choose to be the sole network builder — it was the mandated entity, required by regulation to invest, serve every consumer and maintain reliability. If competition now treats those same sunk, obligation-driven investments as a competitive liability, the policy sends a perverse signal: if today's mandated investment becomes tomorrow's competitive disadvantage, why would any regulated utility invest prudently again?

The fix is to separate the legacy network asset base — recovered transparently from all users — from the incremental network requirement, where a new supplier's added load triggers augmentation costs allocated on a cost-causation basis. This stops existing consumers from subsidising new entrants, while giving new suppliers a real incentive to acquire consumers efficiently rather than opportunistically.

A five-layer architecture worth considering

India could structure this more deliberately than "one DISCOM lets another one in":

1. Network ownership — existing utilities keep their physical assets, on a regulated asset base.

2. Distribution system operation — an independent or functionally independent DSO runs the shared network and polices non-discriminatory access.

3. Competitive supply — multiple licensed suppliers procure power and compete for consumers.

4. Power procurement and legacy PPAs — a transparent mechanism allocates existing contractual liability so the incumbent isn't left holding unrecoverable fixed costs.

5. Universal service — a clearly defined Supplier of Last Resort mechanism, so no consumer is abandoned for being commercially unattractive.

Each layer answers a different question that the current draft leaves implicit: who owns the asset, who operates it neutrally, who competes for the customer, who carries the legacy contracts, and who is responsible when the market itself fails a consumer. Treating these as one undifferentiated bundle — as "shared distribution" tends to be discussed today — is precisely how the cost-allocation problem gets buried instead of solved.

This is a materially more robust model than simply permitting multiple suppliers onto an incumbent's wires.

Someone has to be the Supplier of Last Resort

Electricity isn't a product a consumer can simply do without for a day if their supplier fails. There has to be a Supplier of Last Resort (SoLR) — but that obligation has real cost, and it must be explicitly compensated. Otherwise the asymmetry returns in its starkest form: competitors take the upside in good times, while the incumbent absorbs the downside when the market fails. That's not competition; it's privatised profit and socialised risk.

Shared doesn't have to mean inefficient

There's a risk at the opposite extreme too. Pool network costs without real performance incentives, and you can strip out the very efficiency gains competition was meant to deliver. The DSO needs benchmarking, loss-reduction targets, reliability indices, capital-efficiency norms, time-bound connection standards and performance-linked penalties or incentives — otherwise it becomes just another bureaucratic layer between supplier and consumer rather than a genuinely neutral, efficient platform.

Ten things the framework needs to settle — before, not after, rollout

1. A transparent network-use charge methodology, including return on capital and depreciation.

2. Treatment of historical regulatory assets and liabilities.

3. Allocation of existing PPA fixed costs after consumer migration.

4. A stranded-cost recovery mechanism, where prudently incurred.

5. Cost-causation principles for network augmentation.

6. An independent or functionally independent DSO with real governance safeguards.

7. Non-discriminatory access and transparent network information for all suppliers.

8. A compensated Supplier of Last Resort mechanism.

9. Clear separation of cross-subsidy from network cost, with state subsidies explicitly funded.

10. Rules for consumer switching, metering, data portability and settlement.

Skip these, and "competition" risks becoming a slogan that outruns the economics needed to sustain it.

The consumer test

None of this matters to consumers directly. A household doesn't care whether power reaches it via Network A or Network B — it cares about price, reliability, service, how fast a new connection is made, billing accuracy and whether it can actually switch supplier without friction. If shared networks simply multiply licences while leaving charges opaque and switching cumbersome, the institutional architecture will have changed without the consumer experience changing at all. A regulator that wants to know whether the reform is working should not count licences issued — it should track how many consumers actually switch, how long it takes, and how many complaints arise from the switch itself.

The bigger prize: distribution as a platform

Get the economics right, and the reward is bigger than competitive retail tariffs. A genuinely neutral, well-run shared network could become a platform for rooftop solar, storage, EVs, V2G, demand response, local energy trading and eventually dynamic retail pricing — turning the DISCOM from a one-way electricity deliverer into infrastructure for a much richer energy marketplace. The Draft NEP 2026's DSO and local-trading proposals point exactly there. But a platform only works if its access rules are genuinely neutral — which brings the argument back to where it started.

The real question isn't "how many DISCOMs"

The debate India needs to have is no longer one licensee or many. It's this:

Can India build a neutral distribution network over which multiple suppliers genuinely compete, while ensuring that whoever benefits from that network also carries an appropriate share of its costs and responsibilities?

The government's instinct — sharing infrastructure rather than duplicating it — is right. But sharing the network is only half the job. The other half is allocating its cost fairly. Get that wrong, and India will have simply swapped one problem — monopoly of supply — for another: competition for consumers, paired with an asymmetric allocation of the obligations that make the grid work at all.

The wires can be shared.The responsibilities cannot be.