The Wrong Room: What India's Delimitation Defeat Was Actually Measuring

In April a bill whose sponsors promised the South more seats, not fewer, still fell 54 votes short. That is not a dispute about arithmetic. It is a reading on how much India keeps in Delhi, and on the government nearest its people, which the Constitution created and never funded.

On 17 April 2026 the Lok Sabha voted for a constitutional amendment and defeated it in the same division. The Constitution (131st Amendment) Bill, 2026 would have raised the House's ceiling from 550 seats to 850 and let Parliament decide by ordinary law when the map is redrawn and which census counts. It drew 298 ayes to 230 noes: a majority of total membership, which Article 368 requires, and 54 votes short of two-thirds of the 528 present and voting, which it also requires. The government withdrew the companion Delimitation Bill rather than put it to a vote. The women's quota, law since 2023, lost the delimitation it had been waiting on.

Study what its sponsors said the bill would do. Amit Shah told the House on 16 April that the five southern states held 129 of 543 seats, "approximately 23.76 percent," and would rise to 195 in a chamber half again as large: more seats for everyone, the southern share roughly held. On that description nobody lost anything. The description was not in the bill. Carnegie's post-mortem records that the guarantee "was never codified in the proposed law but was orally communicated in Parliament," while the companion bill pegged the exercise to "the latest published census as on the date of the constitution of the Delimitation Commission," which PRS reads as the 2011 count. Shah's 195 was an aggregate he never broke into states; PRS's modelling of the bill as written gives Tamil Nadu 48 seats in an 813-seat House, nine more than it holds and a fall from 7.2 to 5.9 per cent of the chamber. Members were asked to hand the centre the timing, the census and the map on a promise that lived only in Hansard, and declined.

On this series' reading, the ferocity of an apportionment fight is a diagnostic instrument: it measures how far a federation has failed to devolve. Every argument in April was about who sits in the room. Almost none was about whether the room should be deciding this much. Part III called this democracy's unfinished half. Since 2017 no Indian state has set a tax rate on its own inside the goods and services tax; Article 279A(9) gives the Council a three-fourths threshold and the Union a third of the vote, so nothing carries without it. The Sixteenth Finance Commission then held the states' share of central taxes at 41 per cent against a demand for fifty-fifty. When your revenue is a formula written elsewhere, a seat in the room where the formula is written stops being symbolic. Carnegie puts the malapportionment itself at roughly 8 per cent of Lok Sabha representation, about forty-four of 543 seats: high for a federal democracy, but spread thin across many states rather than pooled in one region as Argentina's is in Buenos Aires. A distortion with no single owner, producing a panic with a very clear one. That gap is the measurement.

India built the institution that would bleed off some of that pressure thirty-three years ago. The panchayati raj ministry's annual report for 2025-26 counts about 2.6 lakh panchayats with 24.24 lakh elected representatives, roughly 49.55 per cent of them women. Then the Reserve Bank looked at the books: about 95 per cent of panchayat revenue arrives as grants from above, and own tax revenue nationally in 2022-23 came to Rs 737 crore. Carnegie, at the other end of the problem, finds a Lok Sabha member representing about 2.7 million people, more than in any other lower house in the world. India's nearest elected government has no money, and its furthest one is unreachable.

The reason is constitutional, and it is one verb used twice. Article 243G provides that a state legislature "may, by law, endow" panchayats with the powers of self-government; Article 243H, that it "may, by law" authorise a panchayat to levy a tax. Reservations shall be made. Finance commissions shall be constituted. Election commissions shall be vested. Everything the tier is made of is mandatory except the two things that would make it a government. That is Part II's ledger logic reaching a modern constitution: what the centre can count, it governs; what it cannot see into the accounts of, it does not release. Part V asks that self-organisation be recognised rather than licensed. Article 243G licenses.

The Finance Commission is the proof, not the villain. It sends local bodies Rs 7,91,493 crore over five years: half the basic grant tied to sanitation and water, the other half untied along with the whole performance component, leaving 60 per cent of basic-plus-performance free against a predecessor that left 40. But twenty of those sixty points are the performance component, which a gram panchayat earns only by raising Rs 1,200 per household a year of its own revenue. The untied money is released by passing a test Delhi set. The gram sabha still resolves which activities the grants fund, so the village picks what gets built. Parliament could not agree who represents Tamil Nadu. Deciding the envelope a gram panchayat spends within, and the revenue it must first raise to qualify, took nobody any trouble.

The strongest objection is that this is cheap to say from outside the fight. Pranab Bardhan argues that where local government is captured, restrictions on its taxing power "may then be desirable," close to the reverse of what is urged here. Rémy Prud'homme's objection is sharper and survives intact: decentralization runs counter to redistribution, and the southern surplus reaches Bihar through the Union budget precisely because the Union holds it. That bounds this argument rather than defeating it. What belongs below is not the transfer but the decision about what the transfer buys. Capture is the real risk, and Ostrom's monitoring and sanctions principles are the conditions that answer it: monitors accountable to the users, sanctions that follow, since a monitor without a sanction is theatre. India constitutionalised that monitor and barely uses it: the gram sabha, the standing assembly of every registered voter in the village. Ambedkar told the Constituent Assembly the village was "a sink of localism, a den of ignorance," and he was right about the failure mode: capture at this tier is usually caste, and the assembly is where it works most directly. Arm the monitor, then, with a sanction and accounts published where the assembly can read them. An empty tier is captured too, and silently.

The Reserve Bank's gradient runs the other way: states with higher devolution show better outcomes in health, education and infrastructure. Say the caveat aloud. The states that devolved most were the best-governed before they devolved. It is correlation, not identification, and the strongest anybody has. Now the hardest fact for this argument. The ministry's Devolution Index for 2024 ranks Karnataka first, Kerala second, Tamil Nadu third: India's best downward devolvers are the states that shouted loudest in April, and none of it took an amendment, a census or Delhi's permission. It took only a legislature willing to use the verb in Article 243G. Read one way that kills the claim. Read another it bounds it. Their devolution is partial and can shrink: Kerala resolved in 1996 to send local government 35 to 40 per cent of its plan and now reports 28.09 per cent. None of it reaches the GST rate or the divisible pool, which is what April was about. The claim is not that a full third tier ends the fight at the centre. It is that a state with somewhere else to put a decision has one fewer decision riding on the room.

State by state, in thirty-day windows running to September, the enumerators are walking India for house-listing; population enumeration follows in February 2027. Those are the figures Article 82 has awaited since it froze the allocation of seats among states on the 1971 census. April bought time to roughly 2027-28, when the argument restarts with real numbers. No published evidence establishes that a funded third tier calms a fight at the centre. Treisman's survey finds the empirical studies "as inconclusive and mutually contradictory as the theories they set out to test," a verdict against confidence in either direction, and Brancati finds decentralization damping ethnic conflict directly while feeding it indirectly through the regional parties it grows. This is an argument, not a forecast, and by the test Part VI set (verify, amend, leave), the panchayat passes one. Scores move, though: on the ministry's own index national devolution has risen from 39.9 to 43.9 per cent since 2013-14. The tier is empty by inertia, and inertia is reversible. A federation that spends the next two years negotiating a better share of one room will hold the same fight in 2028, louder. A federation that spends them filling the tier it already built takes not one rupee back from Delhi. It changes what a seat in Delhi is worth, because fewer of the decisions in a citizen's life route through it. That is a smaller claim than the one the fight is about, and the only one the evidence will carry.