Will the Caste Census Inflate OBC Numbers? The Evidence Says No

The Lift Line

You cannot have a perverse incentive without a reward. Census returns are confidential, so the reward does not exist.

Why This Editorial Matters for Your Exam

Caste enumeration sits at the intersection of social justice, constitutional law and statistical governance. It is among the most examinable GS1 and GS2 topics of this cycle, and this editorial supplies the empirical and legal counter to the most common objection.

GS Paper 1: Salient features of Indian society; social empowerment; caste and social stratification.

GS Paper 2: Mechanisms and institutions for the protection of vulnerable sections; welfare schemes; government policies.

Background and Context

The History of Caste Enumeration

The Constitutional Architecture

The Rohini Commission, constituted in 2017 under Article 340, examined the sub-categorisation of OBCs, on the finding that a small number of dominant communities had captured a disproportionate share of the 27 per cent quota.

The Analysis

1. The incentive argument fails at the first step. A perverse incentive requires a reward. Section 15 of the Census Act, 1948 provides that individual census records are not open to inspection and are not admissible in evidence. A census entry therefore cannot be produced before any authority to claim a benefit. Caste certificates, which actually unlock reservation, are issued by revenue authorities under State-specific procedures involving documentary proof and community verification, a process entirely separate from the Census. Misreporting to the enumerator yields nothing usable.

2. The empirical record contradicts the prediction. National Sample Survey rounds from 1999-2000 to 2017-18 recorded the OBC share within a broadly stable band of roughly 41 to 44 per cent. Reservation benefits existed throughout that period. If self-reported caste were elastic in response to benefit availability, that elasticity should have shown up across nearly two decades of large-sample surveys. It did not.

3. The counter-argument about salience deserves an honest hearing. A decennial Census conducted amid intense national debate is not psychologically identical to a routine statistical survey. Salience can alter response behaviour, and the stability of past survey data does not conclusively rule this out. The right response is not to dismiss the concern but to note that it is testable through post-enumeration surveys, which are a standard census quality-control instrument.

4. The real risk is classification, not candour. India has thousands of caste names, with regional variants, synonyms, differing transliterations and overlapping sub-caste identities. A respondent answering truthfully may give a name that maps to several entries, or to none, in a State list. SECC 2011 is the cautionary precedent: it collected caste data that proved so difficult to classify that it was never fully released in usable form. That was a failure of instrument design, not of respondent honesty.

5. Accurate enumeration is politically consequential regardless of accuracy. If enumeration confirms an OBC share materially above the 27 per cent quota, it will generate pressure to revisit the 50 per cent ceiling set in Indra Sawhney. This is a real consequence, but it is an argument about what to do with data, not an argument against collecting it. Conflating the two is the most common error in this debate.

Data and Institutions Vault

Prelims-grade facts:

• Census Act, 1948; Section 15 makes individual returns confidential and inadmissible in evidence.
• The last full caste enumeration in the decennial Census was 1931.
• Socio Economic and Caste Census (SECC), 2011: caste data never fully released in usable form.
• Mandal Commission: constituted 1979 under B. P. Mandal, reported 1980; 27 per cent OBC reservation implemented 1990.
• Indra Sawhney v. Union of India (1992): upheld 27 per cent OBC reservation, capped total reservation at 50 per cent, mandated exclusion of the creamy layer.
• Article 340 (commission on backward classes); Article 338B (NCBC, constitutional status via the 102nd Amendment, 2018); Article 342A; 105th Amendment, 2021 restored States’ power to identify SEBCs.
• Rohini Commission, constituted 2017, on sub-categorisation of OBCs.
• NSS rounds 1999-2000 to 2017-18: OBC share broadly stable at roughly 41 to 44 per cent.

⚠️ Watch the trap: The Census and the SECC are different exercises under different legal authority. The Census is conducted under the Census Act, 1948 with statutory confidentiality; the SECC 2011 was an executive exercise outside that Act, which is precisely why its data could be considered for release. Do not treat the two as interchangeable.

The Debate

FOR (fears of inflation are unfounded): Section 15 removes any usable reward for misreporting, since census returns cannot establish an entitlement. Two decades of NSS data show a stable OBC share. The identified risk is classification difficulty, which is an instrument-design problem with known solutions.

AGAINST (the concern is not baseless): A politically salient Census differs from a routine survey in ways that historical stability cannot fully settle. More importantly, accurate enumeration will itself generate pressure to breach the Indra Sawhney ceiling, so the exercise is consequential whatever the numbers show.

Balanced verdict: The two sides are largely arguing past each other. The incentive objection, taken on its own terms, is weak: the law removes the reward and the data contradict the prediction. The salience concern is legitimate but testable through post-enumeration verification. The ceiling concern is real but is an argument about policy response, not about data collection. Sound policy requires accurate data, and the case for withholding measurement because the results may be politically inconvenient is not a statistical argument at all.

How to Think About This

When someone predicts strategic misreporting in a data-collection exercise, run a three-step test before accepting it. First, identify the specific reward the misreporter obtains, and check whether the law actually delivers it. Second, look for a natural experiment: has a comparable exercise already been run under similar incentives, and what happened. Third, ask whether the objection is really about the data or about the policy the data might justify.

Applying that test here: the reward does not exist because of Section 15, the natural experiment is two decades of NSS rounds showing stability, and much of the objection is in substance about the fifty per cent ceiling rather than about enumeration accuracy. Separating those three questions is what makes the answer analytical.

Diagram-in-Words

Takeaway Box

Lift line: You cannot have a perverse incentive without a reward. Census returns are confidential, so the reward does not exist.

Prelims hooks: Census Act 1948, Section 15 confidentiality; last full caste census 1931; SECC 2011 caste data never fully released; Mandal Commission 1979, report 1980, implemented 1990; Indra Sawhney (1992) 27 per cent and the 50 per cent ceiling plus creamy layer; Article 340, Article 338B via the 102nd Amendment 2018, Article 342A, 105th Amendment 2021; Rohini Commission 2017; NSS OBC share broadly stable at roughly 41-44 per cent.

Ethics and interview angle: Is there ever a defensible case for a State declining to measure a social characteristic because the results may be politically destabilising?

PYQ linkage: Connects to past UPSC Mains questions on caste as a factor in Indian politics, on reservation policy and the creamy layer, and on the sub-categorisation of OBCs.

Probable question: “Objections to caste enumeration are more often objections to what the data might justify than to the accuracy of the data itself.” Critically examine.

Source: Will the Caste Census Inflate OBC Numbers? The Evidence Says No — Ujiyari.com | Free UPSC & State PCS Editorial Analysis