SignalStack

🌏 The Research Rundown: July 2026

2026-07-08. Published at https://varna.substack.com/p/the-research-rundown-july-2026. This file is a generated copy; the Substack post is the original.

Dear Readers,

Two enumerators may come to your door this summer. One is revising the electoral roll and deciding whether your name stays on the list that lets you vote. The other is listing your house for the Census that will redraw constituencies and reset welfare targets for the next decade.

Both are counting you. Neither is measuring the thing most likely to kill you first: the heat outside.

This issue is about counting. Who does it, whose name comes off, and what the state decides is not worth counting at all. If you have read anything I have been writing lately, you know where my head is: a headline number is a decision about who exists for official purposes, and the arithmetic is the part that hides the decision.

The Rolls: What Actually Happened

The Special Intensive Revision of electoral rolls has gone from a Bihar pilot to something close to a national exercise. On 27 May 2026, the Supreme Court upheld it in Association for Democratic Reforms v. ECI, holding that the Commission’s power to revise flows from Section 21(3) of the Representation of the People Act and Article 324. The same bench also backed the ECI’s decision to exclude ration cards and existing voter ID cards from the documents that establish identity. That matters more than it sounds, because those are precisely the papers a poor household is most likely to hold and least able to supplement with anything else.

Phase III, announced on 14 May, extended the revision to 16 states and 3 Union Territories, leaving only Himachal Pradesh, Jammu and Kashmir, and Ladakh for later. Delhi is the part you can watch live: the door-to-door drive runs from 30 June to 29 July, the draft roll publishes on 5 August, and the final roll follows on 7 October. A name missing from the 5 August draft is off the draft, and getting back on means filing a claim inside a fixed window.

The scale so far, from the completed Phase II (nine states, three UTs, 27 October to 4 December 2025, final lists out by April 2026):

My take: The legal question is closed, and the operational one has taken its place. A revision that places the burden of proof on the citizen will always cut deepest among those who cannot easily prove themselves, and the Court’s decision to disallow ration cards and old voter IDs sharpens that edge rather than softening it. Read the claims-and-objections window after 5 August far more carefully than any raw deletion figure. That window is where a draft number either becomes permanent or gets reversed.

Feature: Deleted

Start with the smallest unit in all of this: one name coming off a list.

When your name drops from a draft roll, nothing announces it. No letter necessarily arrives. You find out when you go to check, or when someone warns you to. Getting back on means filing a claim, producing documents, and hoping it clears before the roll freezes. The design puts the work of proving your own existence on you, inside a fixed window, using paperwork that the people most likely to be cut are least likely to have in order. The Supreme Court has now confirmed that two of the commonest documents a poor household holds, the ration card and the old voter ID, will not, on their own, do the job.

This runs past voting and straight into welfare. A household that cannot satisfy a Booth Level Officer that it belongs at an address tends to be the same household that struggles to hold a ration card, a job card, or a pension record tied to residence and identity. West Bengal has already made the link explicit: its new government has moved to make its flagship cash transfer conditional on proof of citizenship. The money still goes out. What changed is the test you have to pass to reach it.

Look at who fails that test by design rather than by accident. Internal migrants, whom the 2011 Census counted at over 450 million, live between the address on their documents and the address where they actually sleep. Women who move at the time of marriage inherit a paper trail that rarely catches up with them. Renters in cities hold no durable proof of where they live. Older people carry records that predate digitisation. None of these is an unusual case. They are large groups, and the trimming is applied the same way each time, to the same people, which is what separates a patterned outcome from random error.

The official case is accuracy, and accuracy is a real good. Rolls do carry the dead, the migrated, and the duplicated, and a democracy has an interest in a clean one. But accuracy has two failure modes, and the system is built to notice only one. Striking off a genuinely ineligible entry counts as a success. Striking off an eligible citizen counts as nothing, because no one tallies it. When the machinery is set up to celebrate removals and stay silent about wrongful ones, cleaning the roll becomes a one-way ratchet that turns against the people who were already hardest to see.

My take: The word doing the heavy lifting in official language is “purification,” and it should trouble anyone who handles data for a living. You clean a dataset by minimising error in both directions and reporting your false-positive rate. You purify a list by deciding in advance which entries are impure and removing them. Ask for the wrongful-deletion rate rather than the deletion count, and watch how rarely anyone can produce it.

Must-Reads This Month

1. The heat the state does not count Frontiers in Environmental Health, May 2026. Open access. Explainer in Euronews, 10 June 2026.

A new district-level study estimates that a single day of extreme heat is linked to about 3,400 excess deaths across India, and a five-day heatwave to nearly 30,000 excess deaths. Official records put the annual heat toll at 500 to 1,500. The paper covers all 765 districts and counts every excess death during a heat period, not only the cases labelled heatstroke, which is where the official undercount comes from. It also finds that the five states carrying the highest heat mortality account for 66 per cent of national excess deaths while producing only 29 per cent of GDP, so the places least able to pay for cooling and healthcare are the ones dying most. The Euronews write-up is the readable version.

My take: An undercount like this is a budget choice more than a data problem. Heatwaves are still not a notified disaster under the Disaster Management Act, which keeps them outside the relief-funding system a cyclone triggers automatically. Count the deaths honestly, and you create a claim on money. That is exactly why they stay uncounted.

2. India has “virtually eliminated” extreme poverty, on paper, on the Panagariya poverty claim. Free explainer.

Arvind Panagariya, who chairs the Sixteenth Finance Commission, has argued that India has all but eliminated extreme poverty, putting the headcount near 2 per cent. NITI Aayog’s multidimensional measure separately shows a fall from 29.17 per cent in 2013-14 to 11.28 per cent in 2022-23. Both arrive while India still has no official consumption-poverty line updated since 2011-12, so the country is announcing the end of poverty without a settled definition of what it means to be poor.

My take: Pick the line, the price index, and the recall period and you can land almost any poverty number you want, which is what several capable economists are doing in public right now. The defensible claim is smaller: we stopped maintaining the official ruler in 2011 and have argued about substitutes ever since. Wait for the denominator before you celebrate.

3. Representation waits on the count: On the 131st Amendment and delimitation. Free.

The attempt to fast-track women’s reservation collided with the arithmetic of counting. The Constitution (131st Amendment) Bill, 2026, which would have used 2011 Census figures to introduce reservation without delay, was introduced in the Lok Sabha on 16 April 2026 and fell 56 votes short of the two-thirds it needed. So both the delimitation of Lok Sabha seats and the one-third reservation for women now wait on the 2027 Census, whose population count is due in February 2027.

My take: This is why the Census calendar is a political instrument and not a clerical one. Whoever sets the enumeration date sets the delimitation date, and delimitation shifts seats between the south and the north for a generation. A line on a Census schedule moves the federal balance. Watch it accordingly.

Census Watch

Phase I, the House Listing and Housing Census, is being conducted in a staggered manner from 1 April to September 2026, recording houses, assets, and amenities across all 36 states and Union Territories. Population enumeration follows in February 2027. Two things set this round apart from every count since Independence. Caste is being enumerated for the first time since 1931, recording specific jatis rather than the broad SC, ST, and OBC buckets, which hands the reservation and targeting debate a dataset it has spent 95 years improvising without. And self-enumeration through an online portal (se.census.gov.in, available in 16 languages) is allowed for the first time, adding a new source of both convenience and error.

Worth remembering while this runs: the Public Distribution System and the food-security law still allocate on 2011 population figures, so a decade of growth has already pushed millions of eligible people out of the ration net. The new count is what finally corrects that, or locks it in.

Watching for: whether the caste tables are released in full and disaggregated or summarised into uselessness, and how self-enumeration take-up splits across the digital divide, since a portal the connected use and the poor cannot is a bias built in at the front door.

Research Praxis: How to Read a Deletion Rate

A deletion figure with no denominator tells you almost nothing, and yet it is always the number that leads the headline.

When you see “47 lakh electors removed,” run four checks before you react.

Denominator. Forty-seven lakh against Bihar’s 7.89 crore roll is 5.95 per cent. The absolute figure is built to alarm; the ratio is what you actually reason with.

Baseline. Rolls decay on their own through death and migration at a few per cent a year, so a genuine decadal revision should remove a fair slice simply by catching up. The real test is whether removals exceed that plausible churn and which way the errors point.

Draft or final. A name missing from a draft roll can be restored during the claims window, so a draft deletion number is a ceiling, not a verdict. Delhi’s draft lands on 5 August, and its final roll on 7 October; the gap between the two will tell you more than either one alone. Quoting the draft as if it were final is the most common way this statistic is abused.

Error borne by whom. Every verification system trades wrongful removals against wrongful retentions. The politics lies entirely in that trade-off, specifically in whether the citizen or the state has to prove.

Do this once, and no headline deletion count will carry an argument past you on its own again.

Disambiguation: BPL, SECC, MPI, and the Poverty Line

Four terms that get swapped as if they meant the same thing.

The poverty line is a consumption or income threshold, historically set by committees such as the Tendulkar and Rangarajan committees, used to produce a national headcount. India has not had an updated official one since 2011-12.

BPL Below Poverty Line) is an administrative category, and the household lists built from it have long been used to target subsidies. Its lists have carried large inclusion and exclusion errors, which is why holding a BPL card and being poor were never the same thing.

SECC, the Socio-Economic and Caste Census of 2011, used no poverty line at all. It ranked households on deprivation criteria and became the spine of beneficiary targeting for several major schemes.

MPI, the Multidimensional Poverty Index, measures deprivation across health, education and living standards rather than money. NITI Aayog’s national version is built on the NFHS.

Short version: the poverty line counts rupees, the MPI counts deprivations, SECC ranked households, and BPL is the administrative list none of them produced cleanly. Treating them as interchangeable is how the poverty argument stays permanently muddled.

From My Desk

Twice a year I break my own no-self-promotion rule. This is once.

On PolicyGrounds, the thread this quarter has been measurement and money. The column “India Has Never Measured Its Own Inequality” sits squarely on this issue’s theme, and the newer piece on West Bengal, which ties its cash transfer to proof of citizenship, is the welfare mirror of this month’s Feature. If “Deleted” held your attention, that one carries the argument from who gets to vote to who gets paid.

JanVayu, the citizen-run air quality platform, now has an Urban Heat Island section that pairs well with this issue’s heat story, alongside its economic-cost and accountability trackers. With the heat this year, the “how polluted is your ward” view is a good place to see the heat-and-air overlap at a scale you actually live at.

The ImpactMojo Professional Learning Community, the moderated WhatsApp group, remains free across all tiers. It is where the fieldwork questions and the “how do I actually run this” threads land. If you want the working version of these debates, that is the room to be in.

Visual Treat

World Weather Attribution’s rapid study of the 2026 South Asia heatwave is worth opening for the maps alone. The analysis shows how human-caused warming roughly tripled the odds of the April event and made it about 1°C hotter than a pre-industrial climate, and the graphics turn “climate change made this worse” into something you can see. Ten minutes with it and the attribution stops being a slogan.

Funsies

The Census will, for the first time, let you enumerate yourself online. Somewhere, a policy analyst is already dreaming of the day the electoral roll offers the same courtesy so that you can strike yourself off the voter list from home, no Booth Level Officer required. Progress.

Meanwhile, a Member of Parliament has proposed renaming Delhi to Indraprastha. One pictures the enumerators, halfway through the drive, being told to update the address field on 1.45 crore forms. Count first, rename later, is my one methodological request.

You are reading the Research Rundown, a bi-monthly newsletter on development research from South Asia, with a standing interest in who gets counted. If someone forwarded it to you, subscribe here.

Next issue: September 2026.

— Varna

P.S. If your name vanished from a draft roll this year, the claims window is real, and it works more often than the internet suggests. File the claim. Do not take the list’s word for your own existence.