In a historic departure from policy, the Census of 2027 is, for the first time since independence, enumerating castes other than the communities identified as scheduled castes and scheduled tribes. The decision addresses a persistent contradiction in public policy that while caste remains central to affirmative action, welfare and political representation, estimates, especially of other backward classes, are still based on data collected in the 1931 Census.
The colonial censuses from 1872 to 1931 treated caste as a key to understanding and governing India. They recorded and ranked thousands of communities, often converting fluid identities into rigid official categories. Enumeration hardened boundaries, encouraged competition for status and became an instrument of colonial administration. Independent India deliberately broke from that practice, and from 1951, official policy confined caste enumeration to SCs and STs. The exclusion of other castes was, ostensibly, a conscious attempt to foster common citizenship rather than reproduce colonial ethnography.
The drawback of this policy was that it created an evidentiary vacuum. The Kalelkar and Mandal commissions sought to identify socially and educationally backward classes, on the basis of the 1931 Census. Courts have since mandated in several landmark judgements that policies should rest on contemporary, quantifiable evidence.
Current data could, no doubt, strengthen the evaluation of affirmative action, welfare targeting and representation, and replace speculative claims. Disaggregated data by gender and geography (state and district-level) would be especially valuable because caste structures vary sharply across India.
Enumerating caste in a census is, however, not a simple matter of adding a question to the questionnaire. India has thousands of castes, sub-castes, clans, sects and occupational names. Names and identities vary across regions. While central and state OBC lists exist, they do not fully coincide. Unrestricted self-reporting may produce lakhs of variants and unrelated entries. The Socio-Economic and Caste Census (SECC) of 2011 demonstrated that self-declaration without a settled coding framework produced several lakh entries, including spelling variations, surnames, clans, sects and occupational descriptions. Rationalisation of the returns proved exceptionally difficult. Ultimately, the socio-economic data of SECC were released, but the caste data were not.
Religion adds another complication. SC status is presently confined to Hindus, Sikhs and Buddhists, whereas ST status is not restricted by religion. In the case of OBCs, religion has no bearing. In fact, in some states, religious minorities and converts have been accorded OBC status irrespective of their social or economic status.
The key lessons from previous experience are that, firstly, methodology—including the question, coding rules, transliteration and treatment of ambiguous entries—must be finalised and tested before enumeration. Secondly, institutional responsibility must be unambiguous. While caste enumeration should remain within the professional and statutory framework of the census organisation, the use of the data should clearly be the responsibility of the respective line ministries and departments of the state and central governments. Finally, there should be absolute clarity beforehand why the data are being collected, when they will be published and how they will inform policy.
A classification protocol is a basic requirement in this regard. The National Commission for Backward Classes (NCBC), state backward classes commissions (SCBCs) and the Census organisation should jointly develop a harmonised classification framework that must respect state-specific lists while permitting national comparison.
The question, coding system and verification procedure should be tested through pilots in districts representing different linguistic and social conditions. Enumerators should record the respondent’s declaration without interpreting or adjudicating it. Digital tools can assist with transliteration, standardisation, deduplication and the flagging of ambiguous entries, but final classification should remain subject to human and institutional scrutiny.
An expert group comprising census officials, the NCBC, SCBCs, demographers, sociologists, statisticians and constitutional specialists should be constituted before fieldwork. It should supervise classification and post-enumeration processing, publish its coding and aggregation rules, and provide reasons for the treatment of disputed categories. Independent technical audits should test the consistency and integrity of the results.
The NCBC and SCBCs should assume defined responsibility for methodology, state-level verification and policy review. They can subsequently examine representation, welfare access, creamy-layer criteria and under-served groups, without treating population share as the sole test of backwardness.
Need for legislation
An exercise of this sensitivity should not rest only on a Cabinet decision or administrative instructions. Parliament should consider amending the Census Act, 1948, or enacting a supplementary law to specify the purpose, scope, governance and permissible uses of caste data. The law should affirm that the Census records self-declared identity but neither determines backward-class status nor creates an automatic claim to reservation.
The Census Rules should prescribe the question; institutional responsibilities; procedures for coding, verification, aggregation and audit; safeguards against manipulation or disclosure; a publication timetable; and the purposes for which the data may and may not be used.
The confidentiality accorded to census returns must be expressly reaffirmed. Identifiable caste records should not be shared with taxation, policing, benefit-granting or other authorities. Only anonymised, aggregated statistics should be released, backed by penalties for unauthorised access, disclosure or misuse.
Article 338B of the Constitution already empowers the NCBC to investigate safeguards, advise on socio-economic development and recommend measures for backward classes. Rules under the Article or suitable parliamentary legislation should specifically entrust it with advising on methodology, supervising classification and reviewing the implications of the results. State laws governing SCBCs should similarly define their role in verification and state-level analysis.
The law should require the government to place the findings, methodology and an action-taken framework before Parliament within a prescribed period, with state reports placed before their respective legislatures. This would prevent another exercise without a binding framework for classification, publication or use.
Caste enumeration will not by itself deliver social justice. Numbers can reveal inequality, but they can also harden identities and encourage competitive mobilisation. India must therefore count caste not to rank communities, as under colonial rule, but to illuminate disadvantage and strengthen public policy.
Success of enumeration will depend on tested methodology, clearly assigned institutional responsibility, statutory confidentiality, transparent classification, independent audit, assured publication and an agreed framework for policy follow-up. With these safeguards, Census 2027 can definitely provide an empirical basis to policy by replacing conjecture with evidence. Absence of safeguards, runs the real risk of multiplying the confusion on a far larger scale.
C Chandramouli | Former Registrar General and Census Commissioner of India; author of the forthcoming Counting a Nation: The Story of the Indian Census
(Views are personal)
