Census 2027 Records Caste in an Open Column; Centre Yet to Decide How to Tabulate It

The news

New Delhi. More than a month after the Union government notified the questionnaire for the second phase of Census 2027, finalising an open-column method to record caste, officials told The Hindu that no decision has yet been taken on how the government will sort, tabulate and rationalise the data collected. Under the open-column method, a person who is not from the Scheduled Castes (SCs) or Scheduled Tribes (STs) writes his or her caste name freely in a blank field; the person may also choose a “no-caste” option or decline to disclose a caste. Bihar and Telangana, in their caste surveys, instead used a list from which respondents choose. Experts, Opposition parties and associations representing the Other Backward Classes (OBCs) have criticised the choice, arguing that it is bound to repeat the 2011 Socio-Economic Caste Census (SECC), whose open field yielded over 46 lakh different caste names and whose report successive governments shelved. A senior official said those discussions will begin as fieldwork progresses. The second phase, the Population Enumeration (PE) phase, has 40 questions, of which 13 are questions or data fields that were not part of the 2011 census — among them caste, family particulars, ID proofs and COVID vaccinations; the 2011 questionnaire had 29 questions. The caste question is being asked in India’s census forms for the first time since Independence. Officials described the decision path: each division of the Census office works out tabulation for its own questions; the Office of the Registrar General and Census Commissioner of India prepares a holistic proposal; the file goes to the Union Home Secretary; and the final call is taken by the Ministry or the Union Cabinet. Professor G.N. Devy said experts should have been included while finalising the questions, which would have settled doubts about post-collection analysis early. The PE phase has begun in Ladakh and the snow-bound areas of Himachal Pradesh, Uttarakhand and Jammu and Kashmir. In Ladakh, the first State or UT where it is under way, Registrar General Mritunjay Kumar Narayan reviewed the work over four days; an enumerator spends 30 to 60 minutes per household on a mobile application, and the exercise ends by September 30. The syllabus link is GS2 on the constitutional basis of reservation and welfare for vulnerable sections, and GS1 on caste and social empowerment.

Static syllabus linkage

1. The Census is a Union subject run under a 1948 law that guarantees confidentiality. “Census” is Entry 69 of the Union List in the Seventh Schedule, so only Parliament can legislate on it. The Census Act, 1948 empowers the Central Government to notify that a census will be taken, to appoint census officers, and obliges every person to answer the questions put to them; the Census Rules, 1990 lay down the procedure. The exercise is conducted by the Office of the Registrar General and Census Commissioner of India, which sits under the Union Ministry of Home Affairs. Section 15 of the Act makes individual census records confidential: they are not open to inspection and are not admissible as evidence in any civil or criminal proceeding. This is why the Census publishes only aggregated tables, and why the method of tabulation decides what the country actually learns.
2. The SECC 2011 counted caste outside the Census, and its caste tables were never published. The Socio-Economic Caste Census of 2011 was a separate exercise, run by the Ministry of Rural Development in rural areas and the then Ministry of Housing and Urban Poverty Alleviation in urban areas, with the Registrar General providing logistical support; it was not conducted under the Census Act. Its socio-economic data on deprivation was released and has been used to identify beneficiaries of schemes such as PMAY-Gramin and Ayushman Bharat. The raw caste data, which threw up over 46 lakh caste names, spelling variants, clan names and titles among them, was never released. The Union government later told the Supreme Court that the caste data was riddled with errors and unusable for policy. The lesson for 2027 is that an open field without a pre-agreed classification scheme produces a list that cannot be tabulated.
3. Reservation for backward classes rests on three Articles and one ceiling. Article 15(4), inserted by the First Amendment in 1951, allows the State to make special provisions for socially and educationally backward classes, SCs and STs. Article 16(4) permits reservation in public appointments for any backward class that, in the State’s opinion, is not adequately represented in the services. Article 340 empowers the President to appoint a commission to investigate the conditions of backward classes; the Kaka Kalelkar Commission (1953) and the B.P. Mandal Commission (1979) were set up under it. In Indra Sawhney v. Union of India (1992), a nine-judge Bench upheld 27% reservation for OBCs in central jobs, excluded the “creamy layer”, and held that total reservation should not ordinarily exceed 50% save in extraordinary situations. Because the Mandal figure for the OBC share of the population was an estimate from 1931 data, every later debate on quotas has turned on the absence of fresh caste numbers.
4. States have run their own caste surveys, and courts have tested what can be done with them. Bihar conducted a caste-based survey under the State’s executive power and published its findings in October 2023, showing Extremely Backward Classes and Backward Classes together at roughly 63% of the population. The State then raised reservation to 65%, which the Patna High Court struck down in 2024 for breaching the 50% ceiling. Telangana conducted a similar survey using a list of castes. After the 105th Constitutional Amendment (2021), Article 342A(3) confirms that each State may prepare and maintain its own list of socially and educationally backward classes, which can differ from the Central list. A national caste count will therefore have to be mapped against one Central list and many State lists.

Why UPSC loves this

1. GS1 and GS2 both carry this story. The GS1 syllabus asks about social empowerment and the salient features of Indian society, and the GS2 syllabus asks about mechanisms, laws and institutions for the protection of vulnerable sections. Mains questions on reservation have repeatedly turned on the gap between the constitutional promise of adequate representation and the absence of data to measure it. A caste count in the Census is the first time that gap can close, so the method of counting is itself examinable.
2. Prelims tests the institutional plumbing of the Census. UPSC tends to ask which Ministry runs the Census, where the subject sits in the Seventh Schedule, which body conducts it, and how the Census differs from surveys such as the SECC or the National Sample Survey. Article 340 and the commissions set up under it are standard Prelims material. This story gives all of these a current hook.
3. Essay and ethics papers use caste data as a dilemma between recognition and division. Essay topics on identity and on data-driven governance invite the argument that counting caste can either entrench it or expose its inequalities. The open-column choice — self-identification versus a state-defined list — is a concrete illustration of that dilemma for GS4 as well.

Analysis

1. The open column is the honest method; the classification is where the politics hides. Letting people write their own caste respects self-identification and avoids the charge that the State has decided in advance which castes exist, which is the main weakness of the list method used in Bihar and Telangana. But an open field moves every hard choice from the questionnaire to the tabulation room: whether a sub-caste is counted with its parent caste, whether a surname is a caste, whether two spellings are one group. Those are exactly the decisions that the Centre says it has not yet begun to discuss. The critics are right that a method is judged by its output, but wrong to assume the SECC result is inevitable; 46 lakh names is the raw count of entries, not of castes, and it is a coding problem before it is a political one. The counter-view is that a coding problem solved behind closed doors becomes a political problem anyway.
2. Deciding the method after collecting the data invites the suspicion that the result is being chosen. Good statistical practice fixes the classification scheme, the synonym dictionary and the tabulation plan before fieldwork, so that no one can later be accused of arranging the numbers to fit a preferred outcome. Professor Devy’s point that experts should have been consulted at the questionnaire stage is really a point about sequencing. The file path the officials describe — divisions, Registrar General, Home Secretary, Ministry or Cabinet — is an administrative chain with no statutory role for independent statisticians or social scientists. A published tabulation plan and an expert committee with members from outside government would cost little and would protect the Census, one of India’s most trusted datasets, from becoming another contested institution.
3. The count will be read against the 50% ceiling from the day it is published. Indra Sawhney tied reservation to backwardness and adequacy of representation, and later judgments have insisted on quantifiable data before quotas are extended. Caste numbers will therefore be used in two directions at once: by groups arguing that the 50% ceiling should give way to population share, as Bihar tried in 2023, and by those arguing for sub-categorisation within the OBCs so that a few dominant castes do not take most of the benefit. Both uses need tables that map census names onto the Central and State OBC lists. A national count that cannot be mapped onto those lists will satisfy neither side and will leave the courts where they were.
4. The caste question sits in a census that is otherwise more digital and more detailed than ever. The PE phase has 40 questions against 29 in 2011, and enumerators enter data on a mobile application, spending 30 to 60 minutes per household in Ladakh. Digital capture should make tabulation faster, and an app can offer standardised spelling suggestions as a respondent’s caste is typed, reducing the chaos of the SECC. The risk is the opposite: if the app nudges people towards pre-set names, the open column becomes a list method by the back door, without the transparency of a published list. The design of that interface is therefore a policy decision and should be disclosed, not treated as an IT detail.
5. A careful count can reduce, not deepen, caste consciousness — but only if it is used for targeting. The strongest objection to a caste census is that it will harden identities and feed competitive demands for recognition. The reply is that caste already determines access to land, credit, marriage and jobs, and that governments allocate reservations and welfare on caste lines without current data, which rewards the loudest groups rather than the most deprived. Evidence can discipline claims as well as fuel them. The test of Census 2027 will be whether its caste tables are cross-tabulated with education, occupation and assets, so that policy can see deprivation within castes, instead of producing one more population ranking.

Administrator's brainstorm