India does not know how many Other Backward Classes it has. The last national caste count was in 1931, under British colonial administration. Every policy that has been made for OBCs since 1947 rests on nearly century-old data, extrapolations, and educated guesses. The Mandal Commission in 1980 estimated OBCs at 52 percent of India’s population using that colonial-era census and interpolated household surveys. Today, that number is repeated in Parliament, in courts, and in policy documents as though it were established fact. It is not. A caste census would tell us what is actually true. The question India needs to answer is not whether to count, but why it has taken this long, and what holding back the count is really costing.
What the 1931 Census Actually Showed, and Why It Still Matters
The 1931 Census of India, conducted by Herbert Risley’s methodology refined over decades, recorded caste at the granular level. It found that non-upper-caste Hindus, including Scheduled Castes and what we now call OBCs, formed the overwhelming majority of the Indian population. After Independence, Jawaharlal Nehru’s government made the deliberate decision not to enumerate caste in the 1951 Census, arguing that caste-based enumeration would reinforce divisions. Scheduled Castes and Scheduled Tribes continued to be counted because reservation policy required it. OBCs were not.
This created a paradox that has grown more consequential with each decade. The government introduced reservations for OBCs at the central level in 1990 following the Mandal Commission report, covering 27 percent of seats in government jobs and educational institutions. The legal ceiling on reservations was set at 50 percent by the Supreme Court in the 1992 Indra Sawhney case. But the entire framework of who gets how much is built on a demographic estimate that has never been verified with a direct count. Just as India built the world’s largest biometric identity system to establish who its citizens are, a caste census is needed to establish what those citizens need.
The Mandal Commission’s Data Problem
The Mandal Commission submitted its report in 1980. Its central finding, that OBCs constitute 52 percent of India’s population, was derived from the 1931 Census data adjusted for population growth patterns. The Commission identified 3,743 OBC castes and communities across India. Its estimates on the social and educational backwardness of these groups drew on village-level surveys conducted in only 405 villages across the country, a sample that would not satisfy any modern statistical standard for a nation of 700 million people at the time.
The National Sample Survey Organisation, now called the National Statistical Office, has collected data on consumption, education, and employment disaggregated by social group in rounds like the 2004-05 and 2011-12 surveys. These surveys confirm persistent gaps: OBC households had a per capita consumption expenditure that was about 25 percent lower than upper-caste households in rural areas as of 2011-12. But NSSO surveys are household surveys with sample sizes around 100,000 households, not a census of 1.4 billion people. They give useful directional data. They do not give the granular, community-level picture that reservation policy requires.
The Rohini Commission, constituted in 2017 to examine sub-categorisation within the 27 percent OBC quota, found that just 10 communities out of 2,633 OBC communities had cornered 25 percent of all OBC reservations in central government jobs. The top 40 communities had taken nearly 50 percent.
Rohini Commission Interim Report, 2018
This finding, based on actual data from reservation beneficiaries rather than population estimates, shows exactly what happens when policy operates without demographic evidence. Some communities get most of the benefits. Others get almost nothing. Without a census, there is no way to know which communities are over-represented relative to their population and which are systematically excluded.
Bihar 2023: What a State Can Do That the Centre Will Not
In 2023, the Bihar government under Chief Minister Nitish Kumar conducted a state-level caste survey, the first comprehensive count of its kind in independent India. The survey covered all 12.7 crore residents of Bihar across 2.66 crore households. Its findings, released in October 2023, were striking.
- OBCs and Extremely Backward Classes together constituted 63 percent of Bihar’s population, far higher than the Mandal-era estimate of 52 percent nationally.
- Scheduled Castes made up 19.65 percent of the population.
- Scheduled Tribes accounted for 1.68 percent.
- Upper castes, often assumed to be a small minority, were confirmed at 15.52 percent.
- The survey found that 33.58 percent of Bihar’s families lived in kutcha houses, 29.61 percent had no toilet facility, and 94.33 percent of the population earned less than Rs 10,000 per month.
These are not abstract percentages. They are a foundation for targeted policy. If OBCs are 63 percent of Bihar’s population but only 27 percent of central government seats are reserved for OBCs nationally, the data makes a specific argument visible that no amount of political rhetoric can make on its own.
Bihar’s survey was followed by caste surveys in Telangana and Karnataka. Telangana’s survey, released in 2024, found OBCs at approximately 46 percent of the state’s population. Karnataka’s survey reported a similar pattern of OBC concentration in the 50 to 55 percent range. Each state that counts finds a different answer. This is evidence that a national caste census would produce information that changes how policy is made. State-level data-driven governance, similar to how Tamil Nadu used evidence to shape its midday meal programme into one of India’s most impactful social interventions, can transform welfare delivery for OBC communities too.
Central Government Resistance: The Arguments and Their Weaknesses
The central government under successive administrations has resisted a national caste census for reasons that shift depending on who is asking. The stated arguments cluster around three claims.
First: Caste enumeration will deepen caste divisions. This argument was used by Nehru in 1951 and has been repeated ever since. The problem with this argument is that caste divisions are not created by counting them. Caste determines access to education, employment, marriage, and in many parts of India still determines whether a Dalit family can enter a temple or draw water from a village well. The divisions are real. Not counting them does not make them go away. It simply makes it impossible to measure progress or design effective interventions.
Second: A caste census is administratively complex and may be inaccurate. This argument has validity at the margins. Caste identity in India is layered: the same community may be listed under different names in different states. The OBC list itself is different at the central and state levels. But these are problems of data management and standardisation, not fundamental obstacles. The Census machinery that counted 121 crore people in 2011, recorded their religion, language, and disability status, and produced results at the village level, is capable of recording caste. Bihar proved this at the state level.
Third: The data will be misused for political purposes. This is the most honest of the three arguments, and the one no government will say aloud. An accurate OBC count creates political pressure to revise reservation quotas, which has implications for every community and every political party. The BJP, which has historically drawn support from upper-caste voters while expanding its OBC outreach, has particular reasons to avoid a census that might generate new demands from OBC communities. The Congress, which built the current OBC framework through Mandal, now runs the risk that state surveys reveal its own record on OBC representation in party leadership to be poor.
What a Caste Census Would Actually Cost
India’s 2011 Census cost approximately Rs 2,200 crore. The Socio-Economic Caste Census, SECC, which ran alongside it and attempted to collect caste and economic data, cost an additional Rs 3,800 crore. Together, approximately Rs 6,000 crore for the most comprehensive socioeconomic data collection exercise the country has attempted. The SECC data on caste identity was collected but never released by the government, reportedly because it found internal inconsistencies that officials were not confident in. A purpose-designed caste census with clearer methodology would likely cost in a similar range. At current fiscal capacity, this represents approximately 0.02 percent of India’s annual GDP of Rs 295 lakh crore.
For comparison: India spends approximately Rs 1.5 lakh crore annually on OBC-related welfare schemes across central and state budgets, without knowing with any precision how many OBC individuals there are, where they are concentrated, or how well these funds are reaching their intended beneficiaries. A one-time expenditure of Rs 6,000 to Rs 8,000 crore to answer these questions is not a cost. It is an investment in evidence-based governance.
The Supreme Court and the Sub-Classification Question
In August 2024, a seven-judge Constitution Bench of the Supreme Court of India ruled in the Punjab vs. Davinder Singh case that states can sub-classify OBCs within the 27 percent reservation quota to give priority to the most backward communities within the OBC category. The ruling overturned a 2004 judgment and opened the door to a more granular reservation framework that had been sought by EBC communities who felt that dominant OBC castes captured most benefits.
This judgment makes a national caste census more urgent, not less. Sub-classification requires knowing which communities are most backward relative to others. You cannot rank communities by backwardness without data on their current educational attainment, income levels, and representation in public employment. The Court has now sanctioned a policy architecture that requires exactly the kind of data a caste census would provide. Without it, sub-classification becomes another political negotiation rather than a data-driven exercise.
What Mandal II Would Need to Look Like
If a national caste census were conducted with the 2026 Census, which the government has already delayed from 2021 and has not announced a new date for, the data would enable a genuine recalibration of the OBC reservation framework. This is what is sometimes called Mandal II, a reference to a potential second Commission-level review of OBC policy.
A Mandal II exercise with actual census data could address several specific failures of the current framework:
- The 50 percent ceiling set by the Indra Sawhney judgment is under challenge in Tamil Nadu, which has 69 percent reservations under a state law. If OBCs are genuinely 60 percent or more of the population rather than 52 percent, the constitutional basis for the ceiling and the distribution within it require re-examination. The Court in Indra Sawhney was explicit that the ceiling was not absolute and could be exceeded in extraordinary circumstances.
- The central OBC list has not been comprehensively revised since the Mandal Commission. Several communities argue they have been wrongly excluded. Others argue that communities that have achieved upward mobility should graduate out of the OBC category. Without current data, these arguments cannot be resolved fairly.
- Women from OBC communities face compounded disadvantage. NSSO data consistently shows that OBC women have lower labour force participation rates, lower literacy levels, and more restricted access to healthcare than OBC men. A caste census that disaggregates by gender could enable gender-sensitive OBC policy, which currently does not exist in any systematic form.
The Real Cost of Not Counting
The 2021 Census was due in February 2021 and was postponed because of the COVID-19 pandemic. It has not been rescheduled. India now heads into a general planning cycle for its Fifteenth Finance Commission recommendations without current population data. The delay in the Census itself has consequences across health planning, school infrastructure allocation, parliamentary delimitation, and urban development budgets.
Adding a caste module to the overdue Census is not a complication. It is an opportunity that, if the government misses it again, will not return for another decade. Each year without a caste count is a year in which approximately Rs 1.5 lakh crore in welfare spending is allocated without knowing whether it is reaching the right people. Each year without data is a year in which courts are asked to rule on sub-classification, creamy layer exclusion, and the adequacy of quotas without the empirical foundation that would make these judgments rational rather than arbitrary.
The Mandal Commission used 1931 data in 1980, forty-nine years after the last count. If India conducts a caste census in 2026, it will have been using 1931 data for ninety-five years. Some policies make sense as temporary measures pending better information. When the temporary measure stretches across a century, the absence of information is itself a choice, and a costly one.
Counting as a Democratic Act
A democratic state allocates resources based on need. Need requires evidence. India has more than a billion people, a constitutional commitment to social justice, and a welfare architecture that affects hundreds of millions of lives. The argument that counting these lives by caste will somehow make caste worse is the argument of a government that prefers comfortable ignorance to uncomfortable knowledge.
Bihar counted. Telangana counted. Karnataka counted. Each state that has done so has found that the data does not cause riots. It causes policy arguments, which is exactly what a democracy is supposed to do with evidence about inequality.
A national caste census is not a left-wing or right-wing demand. It is a demand for the basic data infrastructure that any serious social policy requires. The question of what to do with the data, whether to raise or lower quotas, how to sub-classify, where to draw the creamy layer, is legitimately contested. The question of whether to collect the data at all has only one defensible answer.
What You Can Do
The demand for a national caste census is one that cuts across political parties, though it is more consistently supported by opposition parties and OBC political formations. If you believe evidence-based policy matters, you can support this demand by engaging with it in its specific, factual terms: India has not counted its OBCs since 1931, the Mandal Commission estimate of 52 percent has never been verified, and the Bihar survey suggests the true figure is likely higher. These are not ideological positions. They are facts about what India does and does not know about itself.
The 2026 Census, when it finally happens, is an opportunity that will not come again for a generation. The case for including a caste module has been made in courts, in commissions, and in state surveys. What remains is political will.