Sixteen years without a census is not an administrative failure. It is a policy choice — one that has quietly determined who eats, who is represented, and whose existence the state acknowledges.

In the spring of 2020, I was in the middle of fieldwork examining how seasonal migrants from Bihar's Sitamarhi district navigated welfare entitlements in Delhi. The men I spoke with — construction workers, vegetable vendors, domestic staff — had come to the capital in the years since 2011, the last year India had counted its population. They did not appear in any ration list. Their names were not on any housing scheme beneficiary roll. The neighbourhood they lived in — a dense cluster of ten-by-ten rooms in Uttam Nagar — was invisible to the municipal authority's ward planning documents, which still reflected boundaries drawn from a survey conducted when most of these men had not yet arrived. They existed as people, evidently. They did not exist as data subjects.

This essay is about the Census 2027 — India's first population enumeration since 2011, Phase 1 of which officially launched on April 1, 2026 — and what the sixteen-year gap between these two exercises has actually meant for governance and welfare. The specific provisions matter: the digital-first implementation, the caste enumeration question, the delimitation it will trigger, and the methodological choices still unannounced. But the deeper question is structural. A state that does not count its people has made a choice about who counts. The census delay was not a bureaucratic accident; it was the outcome of five intersecting political calculations, each of which benefited from not knowing, and each of which is now being forced — incompletely, partially, and controversially — to confront what the numbers actually say.

What Phase 1 involves, and what it has not yet settled

The Houselisting and Housing Census began on April 1, 2026, and runs through September 30, across 36 states and union territories, 7,092 sub-districts, and approximately 640,000 villages. Administrative boundaries were frozen on January 1, 2026. President Droupadi Murmu, Vice President C.P. Radhakrishnan, and Prime Minister Modi completed their self-enumeration online on the day of launch.

This is India's first fully digital census. No paper forms will be used at any stage. Enumerators will use a dedicated mobile application — Android and iOS, functional offline — that uploads data when connectivity is restored. Citizens can self-enumerate through se.census.gov.in in 16 languages, receiving a 16-digit unique ID that enumerators will verify on their doorstep visit. The Census Management and Monitoring System provides real-time tracking dashboards. A pilot covering approximately 5,000 census blocks ran in November 2025. The full exercise will deploy 3.1 million enumerators and supervisors across roughly 80,000 training batches — an operation generating an estimated 10.2 million person-days of work alone.

The Phase 1 questionnaire was gazetted on January 22, 2026, by Registrar General Mritunjay Kumar Narayan. It's 33 questions cover building materials, household amenities, assets, and — new in this round — the main cereal consumed by the household, a proxy for food security that was absent in 2011. Phase 2, the population enumeration itself, is scheduled for February 2027 with a reference date of March 1, 2027, and will collect individual-level data on age, religion, education, occupation, migration, fertility, and — for the first time since 1931 — caste.

The Union Cabinet approved a total outlay of ₹11,718.24 crore on December 12, 2025, with ₹6,000 crore allocated in the 2026–27 Union Budget — a sixfold increase from the prior year. The 2011 Census cost approximately ₹2,200 crore; the Registrar General had initially requested ₹14,600 crore. The Cabinet approved ₹2,882.84 crore less than requested, a gap that analysts have noted without official explanation.

Two critical questions remain unanswered. The first is the caste enumeration methodology. The Cabinet Committee on Political Affairs approved caste enumeration on April 30, 2025, but the methodology has not been made public. Whether respondents will self-declare their jati against an open text field — the approach that produced 46.73 lakh distinct caste and sub-caste entries in the SECC 2011 and generated 82 million data errors — or select from a pre-vetted state-specific list, as Bihar's successful 2023 survey used, will determine whether the data is usable. The Supreme Court has already flagged that "mere self-declaration may not ensure accuracy." The second is the National Population Register. Unlike the original Census 2021 plan, which allocated ₹3,941 crore separately for NPR updating, Census 2027 contains no NPR budget. The government has stated it has "not yet decided" on NPR. This matters enormously for communities — particularly Muslims, Adivasis, and stateless persons in border districts — who feared the original NPR-NRC pipeline.

One hundred and twenty million people who do not exist on the ration list

The most immediate human cost of the census delay is arithmetically straightforward, even if its political and developmental implications are anything but. The National Food Security Act (2013) mandates coverage of 75% of the rural population and 50% of the urban population, calculated from the most recent census. The most recent census was in 2011. The statutory beneficiary ceiling has therefore been frozen at approximately 813 million people — a number set in stone as India's population grew by approximately 250 million.

Development economist Jean Drèze estimated in 2025 that applying the NFSA's coverage ratios to the National Commission on Population's current projections would require coverage of over 920 million people. With actual coverage at approximately 806 million, he concluded that over 120 million people "may be regarded as unfairly excluded due to the Census delay." Earlier estimates by Drèze, Reetika Khera, and Meghana Mungikar put the figure at 100 million in 2020; a Down to Earth calculation using a projected 2025 population of 1.417 billion found 137–151 million excluded. The Supreme Court directed the Centre on June 29, 2021, to re-determine beneficiaries beyond 2011 caps. The government's affidavit responded that revision would be "possible only after the relevant data of the next population Census is published."

The distribution of this exclusion follows familiar patterns of Indian disadvantage. Uttar Pradesh faces an estimated 28 million excluded individuals, and Bihar approximately 18 million. These are disproportionately SC, ST, and OBC households. A 2024 study by Shrinivas, Baylis, and Crost found that PDS expansion under NFSA prevented 1.8 million children under five from being stunted — making the exclusion not an accounting anomaly but a nutritional emergency with measurable consequences for the next generation. The FAO estimates that between 2021 and 2023, 194 million Indians — 14% of the population — were undernourished.

The data freeze extends across the entire welfare architecture. The Socio-Economic and Caste Census (SECC) of 2011 — the last comprehensive socio-economic enumeration — remains the basis for identifying beneficiaries for PM Ujjwala Yojana, PM Awas Yojana, and Ayushman Bharat. People who fell into poverty after 2011 — through demonetisation, pandemic-related loss of livelihood, or economic displacement — are not represented in these databases. Women who married and moved households after 2011 face particular exclusion, since the SECC typically recorded the household of origin. Children born in the subsequent 15 years have no formal place in the welfare-targeting architecture built on that survey.

This is one of the central arguments of a previous essay on welfare architecture that I published, examining what ails India's health system: the government's preference for insurance-based coverage relies on a beneficiary identification infrastructure that does not exist. You cannot enrol people in Ayushman Bharat if you cannot identify them. The SECC 2011 — which the government's own affidavit to the Supreme Court described as "inaccurate and unusable for any purpose whatsoever" for its caste data — is the document on which this architecture rests. The census delay has not merely left welfare underfunded. It has left welfare misallocated.

Nine postponements and the reasons behind them

India had not broken its decennial census pattern since 1881. The 1941 census was conducted during World War II. The 1971 census was conducted during the Bangladesh Liberation War. Between March 2020 and late 2024, the administrative boundary freeze deadline was extended nine times in a row, each time quietly deferring the exercise. Understanding why requires examining each political fault line that made a census politically costly.

The COVID-19 explanation was partially legitimate: deploying 3 million enumerators — predominantly primary school teachers — during a pandemic posed real logistical and health risks, and 2020 and 2021 were genuinely disrupted years. But the continuation of the delay through 2022, 2023, and 2024 requires a different account. By mid-2022, India had vaccinated hundreds of millions. Bangladesh conducted its census in June 2022. Nepal completed its in November 2021. The United States proceeded in April 2020, at the peak of the pandemic. The government's own statement to the Lok Sabha cited "COVID and its aftermath" — a formulation whose elasticity was not lost on demographers or political analysts.

The NPR-NRC controversy was the first structural obstacle. Census 2021 was designed to run simultaneously with NPR updating, which the Registrar General's office had described as "the first step towards creation of the National Register of Indian Citizens." In December 2019, as the Cabinet approved ₹3,941 crore for the NPR exercise, the Citizenship Amendment Act triggered nationwide protests. Ten states and two union territories passed Assembly resolutions opposing the NPR update. West Bengal and Kerala imposed a complete hold on NPR-related activities, threatening the census itself with a credibility problem: if minority communities, uncertain about citizenship implications, misreported information to enumerators, the data would be contaminated at the source. A senior BJP leader told ThePrint at the time that the calculation was explicitly electoral — not about creating a "fresh political storm before 2024."

The caste enumeration question was the second fault line and, arguably, the most consequential. India has not recorded OBC caste data in a Census since 1931. Opposition parties across the political spectrum — Congress, RJD, JDU, Samajwadi Party — repeatedly demanded caste enumeration. The BJP's stated objection was that caste data would "heighten social tensions." The unstated calculation was more substantive: transparent OBC population data would put the Supreme Court's 50% reservation ceiling under severe political pressure; expose the structural dominance of upper castes in public-sector institutions; and fracture the cross-caste Hindu identity that Hindutva politics requires. When the BJP reversed this position on April 30, 2025, it was not a principled reconsideration. The party had just lost 63 seats in the 2024 Lok Sabha elections and watched Congress campaigns built around caste census demands gain traction across SEBC-heavy constituencies.

A third factor, documented by The Lancet and subsequently by JOGH, was exposure to COVID mortality. Research published in the Journal of Global Health in 2025 calculated that India's COVID-19 excess death ratio was 7.2 to 10.7 times the officially reported figure, with state-level ratios ranging from under 2x in Kerala to over 40x in Gujarat. A comprehensive civil registration data collection — running concurrently with census operations — risked surfacing the precise demographic evidence of this discrepancy in ways that would have been politically irrecoverable in an election year.

Finally, the delay served a structural governance function that none of the above explanations captures, but that Pronab Sen, former Chief Statistician, identified directly: "We are shooting in the dark. We do not know how many people are poor in India and how many have transitioned from the BPL to the above-poverty-line category in the last 12 years." Frozen welfare lists maintained the existing distribution of entitlements without requiring fiscal expansion. The 120 million unregistered poor cost the government nothing — because the government had no data confirming they existed.

Bihar cracked open what Delhi withheld

The most consequential challenge to the census delay came not from Delhi but from Patna. The Bihar caste survey, released on October 2, 2023, was India's first comprehensive state-level caste enumeration in decades, and it immediately reshaped national politics in ways the central government had spent years trying to avoid.

Conducted in two phases using 304,000 enumerators and a dedicated mobile application called "Bijaga," the survey covered 130.7 million people across 38 districts. Its headline findings were explosive: OBCs and Extremely Backward Classes combined constituted 63.14% of Bihar's population — far exceeding their 27% share of central reservation quotas. Scheduled Castes comprised 19.65%, Scheduled Tribes 1.68%, and the unreserved general category just 15.52%. Brahmins — consistently overrepresented in elite institutions, civil services, and the upper judiciary — accounted for 3.65% of the population. The economic data was equally stark: 34.13% of families reported monthly incomes at or below ₹6,000. Only 7% held graduate degrees. Less than 1% owned a four-wheeled vehicle.

The methodological lesson was as significant as the political one. Bihar's survey used a pre-vetted list of 214 state-specific castes, unlike the open-ended text field that produced SECC 2011's unusable output. The Rohini Commission — a government body constituted in 2017 to examine OBC sub-categorisation — submitted a roughly 1,000-page report in July 2023 finding that 97% of centrally reserved OBC positions went to just 25% of OBC sub-castes, while 983 OBC communities — roughly 37% of all OBC sub-castes in the central list — had received zero representation in central government employment over three decades. That report remained in "cold storage" as of early 2026. Bihar's survey made its suppression politically impossible to sustain.

The political fallout was immediate and nationwide. The Bihar government attempted to raise caste-based reservation quotas from 50% to 65% — bringing total reservations, including the 10% EWS quota, to 75% — but the Patna High Court struck it down on June 20, 2024, for exceeding the Supreme Court's 50% ceiling. Congress made a national caste census central to the 2024 election campaign. PM Modi, on the day Bihar's results were released, accused the opposition of "dividing the country along caste lines." The BJP ultimately approved caste enumeration in April 2025 — a reversal that sheds more light on the party's electoral arithmetic than on any principled reckoning with the demand.

The delimitation crisis and the federal compact under strain

Census 2027's second major political consequence is parliamentary delimitation — the redrawing of constituency boundaries that have been frozen since 1976. The 42nd Constitutional Amendment froze the allocation of Lok Sabha seats at the 1971 Census figures to avoid penalising states that controlled population growth. The 84th Amendment (2002) extended this freeze until "the first Census taken after the year 2026." That census is now underway, and the freeze is about to lift.

The demographic divergence over five decades is severe. Southern states achieved below-replacement fertility rates — Tamil Nadu at 1.8, Kerala at 1.8 — while northern states lagged significantly, with Bihar currently at 3.0. A purely population-proportional reallocation of 543 seats would transfer roughly 32 seats from the south to the north: Tamil Nadu losing 8, Kerala 8, and the two Telugu states 8, while Uttar Pradesh gains 11 and Bihar gains 10. This is the core of what The Week, in a February 2026 cover story, called "the delimitation dilemma: why India's north-south divide is at a breaking point." Tamil Nadu contributes approximately 9% of India's GDP with roughly 6% of its population. Kerala has 99% literacy. The southern states collectively contribute far more to the national revenue pool than their population share would suggest — and delimitation based purely on headcount would reduce their parliamentary leverage over that very resource allocation.

The government's response, announced on April 8, 2026, is expansion rather than redistribution. A Constitution (131st Amendment) Bill proposes expanding the Lok Sabha from 543 to 816 seats on a 50% pro-rata basis using 2011 Census data, with no state losing any seats in absolute terms. Under this formula, Uttar Pradesh rises from 80 to 120 seats, Bihar from 40 to 60, while Tamil Nadu goes from 39 to approximately 58 and Kerala from 20 to 30. The bills are to be introduced in a special parliamentary session from April 16–18, 2026.

The opposition's objection is both constitutional and federal. Karnataka Chief Minister Siddaramaiah calculated that five southern states would gain 63–66 seats under the expansion formula, while seven BJP-dominated northern and western states would gain 128–131 seats, nearly double. The absolute parliamentary gap between Uttar Pradesh and Kerala widens from 60 to 90 seats. Congress general secretary Jairam Ramesh described the expansion as a "Weapon of Mass Distraction," and P. Chidambaram noted the session timing — coinciding with Tamil Nadu and West Bengal state elections — would exclude 67 opposition MPs from voting on constitutional amendments. Tamil Nadu Chief Minister M.K. Stalin has warned the DMK "will not stand by and watch any attempt that places the rights of southern states at stake." Former Chief Election Commissioner S.Y. Qureshi has questioned whether delimitation is necessary for women's reservation at all — the Act, as written, does not require a fresh delimitation exercise as a precondition.

The Nari Shakti Vandan Adhiniyam (106th Constitutional Amendment) — the women's reservation law passed in 2023 — provides 33% of Lok Sabha seats for women, but only after delimitation. The government's new bills target the 2029 elections: of 816 seats, 273 would be reserved for women, including sub-reservations for SC and ST women. OBC women's reservation has not been included, despite the Bihar survey demonstrating that OBC women face compounded disadvantage. The pattern replicates the SECC 2011's failure at a constitutional level: a framework that promises inclusion while designing around the communities most in need of it.

A digital census in a country where 27% of rural users can navigate a government website

Census 2027's fully digital architecture is an administrative experiment with no precedent at this scale in any country with an infrastructure profile like India's. Internet penetration stands at roughly 65% nationally, but active rural usage is only 37%, and just 27% of rural users are digitally literate in any functional sense. Bihar has 25% internet penetration. The self-enumeration portal assumes access to a smartphone, a mobile number, and the ability to navigate a multilingual government website — conditions that exclude the populations disproportionately most at risk of undercounting: rural poor households, lower-caste communities, the elderly, and the mobile labour force of seasonal migrants.

The hybrid model — combining self-enumeration with door-to-door enumerator visits — is intended to close this gap. Enumerators will visit every household regardless of self-enumeration status. But enumerators are expected to use their personal smartphones, leading to variability in device quality, application stability, and data security. India's 2025 pilot covered 5,000 census blocks out of approximately 640,000 total — less than 1% of the operational footprint. No country has ever run a fully digital census of a population of this size and with this level of geographic and linguistic diversity.

India's 16-year census gap places it in the company of historical anomalies. Lebanon has not conducted a census since 1932. Among democracies, India's gap is without parallel. Nordic countries have moved to register-based censuses using linked administrative data, producing annual population statistics without field enumeration. France operates a rolling census covering the entire population every five years. These systems require decades of administrative infrastructure development — precisely the investment India has not made. The Sample Registration System, the National Family Health Survey, and every NSO survey conducted since 2011 have used Census 2011 as their sampling frame. Every estimate, every ratio, every district-level projection of the past fifteen years has been extrapolated from a population snapshot taken in 2011. The errors compound.

The 16th Finance Commission report, submitted in February 2026 and governing tax devolution to states until 2031, used 2011 Census data with population assigned a 17.5% weight in horizontal allocation. The National Health Mission plans district-level health infrastructure based on population projections that have historically shown 2–6.5% errors at the state level within a single intercensal decade. The UDISE+ education planning system computes enrolment ratios against "moderated official child population" derived from those same projections; as the late NIEPA professor Arun C. Mehta documented, the absence of the 2021 Census made it "rather difficult to correctly project child population" for any planning purpose. The migration data vacuum may be the most consequential for economic policy: Census 2011 recorded 450 million internal migrants — 37% of the population. Current estimates suggest over 600 million. The Census remains India's only comprehensive source of migration data, and without it, the One Nation One Ration Card scheme, portable health entitlements, and urban infrastructure planning operate on structural guesswork.

This is the deeper argument that Pronab Sen, former Chief Statistician, has consistently pressed: "There is frankly no alternative to the census for a country like India. If the census data is too old, then the survey estimates are no longer reliable." The Carnegie Endowment for International Peace's 2023 assessment of India's statistical system described the census delay as producing "a data ecosystem in which the primary sampling frame is increasingly unrepresentative of the population it claims to describe."

What would a better census architecture require

The Census 2027 exercise, when completed, will generate the most comprehensive population dataset India has ever had — including caste data that has been absent for nearly a century. But getting there adequately requires several decisions that have not yet been made, and structural investments that go well beyond this census cycle.

The caste enumeration methodology must be published and consulted upon before Phase 2 begins in February 2027. The difference between an open-ended text field and a pre-vetted list is the difference between data and noise. Bihar's 2023 survey demonstrated that a pre-vetted methodology is both feasible and credible. The Rohini Commission's suppressed report on OBC sub-categorisation should be released before the census question design is finalised, since that report's findings on which communities have historically been excluded should shape the question architecture. The Supreme Court's observations about self-declaration accuracy deserve a formal methodological response from the Registrar General's office — not a deferred acknowledgement.

The risks of digital exclusion require a specific operational mitigation strategy. Enumerators in low-connectivity, low-literacy districts should be equipped with standardised devices rather than personal phones. The pilot should be expanded to at least 10% coverage before Phase 2 deployment. An independent technical committee, with civil society and academic representation, should be given real-time access to data quality monitoring dashboards and the authority to flag anomalies for operational correction.

The welfare entitlement update must not wait for the final publication of census data. The Supreme Court's 2021 direction to update beneficiary counts remains unimplemented. The government could use interim state-level demographic estimates — which the Registrar General's office maintains — to provisionally expand PDS coverage now, with final recalibration once 2027 data is published. There is no technical barrier to this. The 120 million excluded from food rations are not excluded because the numbers are unavailable; they are excluded because the government has chosen to use 2011 numbers.

The delimitation process should be conducted through a genuinely consultative Delimitation Commission with representation from all affected states, rather than being settled by constitutional amendments tabled in a four-day special session from which the governments of major opposition states are structurally excluded. The Commission's terms of reference should explicitly address the demographic penalty on fertility-controlling states — either through the expansion formula, differential seat weightings that reflect fiscal contributions, or a compensatory Finance Commission adjustment. These are difficult political negotiations. The difficulty is not a reason to avoid them.

Finally, India needs an inter-censal statistical infrastructure investment that does not require every planning decision to wait 15 years between data updates. The Periodic Labour Force Survey, NFHS, and Annual Health Survey are partial substitutes but not census equivalents, and they cannot provide the district-level granularity that health, education, and welfare planning require. Administrative registers — civil registration of births and deaths, UIDAI's Aadhaar database, state ration lists — could be linked and standardised to produce running population estimates between census cycles. This would require interoperability among three systems that currently operate under different legal frameworks. The current census could be the occasion to begin designing that architecture, even if the infrastructure does not exist to deploy it immediately.

The men in Uttam Nagar, and the count that is coming

I do not know whether the men I met in Uttam Nagar in 2020 are still there. Many seasonal migrants returned to their villages during the lockdowns and never came back to the city. Some undoubtedly did return. Their ration cards, if they were ever issued, remain registered at their villages of origin. Their addresses in Delhi were never formally recorded. Their children, if they were born in the capital, may have been counted in school enrolment registers but not in any welfare identification document.

These are not marginal cases. The World Economic Forum estimates India has 139 million internal migrants as of 2017; current projections suggest significantly more. The India Forum's analysis of migrant exclusion characterises their invisibility as structural rather than incidental: the design of India's welfare architecture presupposes a sedentary, village-resident population in ways that systematically exclude mobile labour. The Census 2027, if its migration questions are well-designed and its data quality mechanisms are robust, could provide the first comprehensive picture of this population in over a decade — and the evidence base for finally building a portable entitlement system that people can carry with them when they move.

The sixteen-year gap was not accidental. It was the product of a government calculating that the risks of knowing — the NPR protests, the caste data exposure, the mortality figures, the fiscal pressure of updating welfare lists — outweighed the risks of not knowing. Those risks were borne overwhelmingly by the 120 million people who could not access rations, the migrants whose children were planned for by health centres that did not know they existed, the OBC sub-communities whose exclusion from reserved positions the Rohini Commission documented and the government suppressed, and the southern states whose fiscal contribution to the national common pool was calibrated by a parliament frozen in 1971.

The count has begun. The data will take years to publish in usable form. The delimitation fight is just starting. The caste methodology has not been disclosed. The welfare update still has not happened. The Census 2027 is necessary but not sufficient — and whether it redeems the sixteen years it has taken to arrive depends on decisions that have not yet been made, about questions that the government has consistently preferred not to answer.

Further Reading

On Census 2027 operations and the delay

PIB: Registrar General addresses press conference on Census 2027, March 30, 2026

Al Jazeera: India's Census 2027 — why it is controversial, April 1, 2026

Down to Earth: India's first digital census — concerns about delays and privacy

NUS ISAS: India's delayed census — why it matters

Indiaspend: Census delays leave 120 million Indians without access to foodgrain

Newslaundry: Census delay cuts access to subsidised foodgrain for 120 million Indians

On caste enumeration and the Bihar survey

The Wire: A look back at India's caste census journey

Business Standard: Rohini Commission decoded — understanding OBC sub-categorisation

Harvard: Caste Census Data for a Just Republic (Surinder S. Jodhka and Ghanshyam Shah)

The Lancet: India's indefinitely delayed census (2023)

On delimitation and federal politics

The Week: The delimitation dilemma — why India's north-south divide is at a breaking point

The India Forum: India's delimitation dilemma — challenges and consequences (Mohd. Sanjeer Alam)

ThePrint: How census-based delimitation could shake up politics and disadvantage the south

On welfare targeting and data infrastructure

IDR: Delayed census — how India's welfare schemes are suffering

Down to Earth: Control and delete — lack of data updating has left millions of poor Indians hungry

The India Forum: India's migrants — exclusion by design

SPRF: New India's policies from old India's data — the 16th census delay and its impact