‘Census 2027 Should Enumerate Caste Power Alongside Caste Apartheid’

What triggered your interest in the politics of caste data? 

I first became familiar with the limitations (and possibilities) of census data during my graduate studies in sociology, when co-authoring a paper that used Census 2001 data to study the segregation by caste/tribe in Indian cities. My dissertation proposal following this research coincided with the aftermath of the political leadership agreeing to collect caste-wise data in Census 2011. The 2010 public concession [the then-Manmohan Singh-led government agreed to collect caste-wise data in the census] appeared to be a major shift in census policy and created the chance to study a historic census that could provide a window into knowledge-making processes, caste inequalities, and state power. 

I didn’t anticipate that the state project  would continue for years and fail – with political leaders eventually deciding not to publish the collected caste-wise data.

 The writings of Jotirao Phule and BR Ambedkar helped me to make sense of the observational, interviews, and document data that I collected between 2011 and 2016 on the first caste-wise enumeration in independent India. These 80- to 150-year-old texts, along with more recent research by historians and social scientists, such as Satish Deshpande, Christophe Jafferlot, Gail Omvedt, Shailaja Paik, V. Geetha, Gopal Guru, and S.V. Rajadurai, among many others, were crucial for me to develop a historically grounded understanding of the present. They also offered a theoretical framework for why a nationwide survey, initially lauded for its use of technology, failed to publish caste-wise data.

Between 1931 and the 1990s, which political sparks fired the discourse on caste census?

After a caste-wise enumeration was removed from Census 1951 (and justified as a policy break from colonial practices), the first Backward Classes Commission (BCC) in its 1955 report called for a full caste-wise enumeration as part of Census 1961. The BCC argued on ideological grounds: if the government was to play an active role in destroying caste, full information on caste was required at regular intervals. The political leadership at the time dismissed the 1955 report, and the BCC chair (Kaka Kalelkar) even repudiated the contents of the report that he had signed. 

Between the 1960s and 1980s, decentralised demands for a caste census occurred alongside the political mobilisation of Other Backward Classes (OBCs), and often in relation to legal cases concerning reservation programmes. The political leadership effectively deflected the issue of collecting caste-wise data to state governments, who were asked to set up and expand state-level agencies and commissions to address the needs of OBCs. Many of the same state-level agencies started to demand a full caste count in the census. In Mysore, for example, the first BCC headed by R Nagana Gowda in 1960 strenuously complained of insufficient data, and urged the state government to pressure the center to collect detailed caste-wise data in Census 1961.

Twenty-four years after the first report, the second BCC (chaired by BP Mandal) also advocated for a full caste count (Mandal wrote three letters to successive home ministers between June 1970 to March 1980 requesting a caste-wise enumeration in Census 1981). Without updated census data, the BCC made its estimates based on projections from 1931 census data—and recommended a 27% reservation quota that was equal to half the population estimate of OBCs. After political leaders extended reservations to OBCs in 1990 and the policy faced legal scrutiny, both those for and against the Mandal Commission’s recommendations called for more data on caste.

What has been the political history of census making since the 1990s?

The post-Mandal world exposed the bureaucratic resistance to enumerating caste in the census. There was an administrative need for data on castes, and the executive bureaucracy should have minimally expanded census data collection to enumerate OBCs following the expansion of reservations, but that didn’t happen in Census 2001. At the time a BJP-led coalition government was in power, and the census commissioner M. Vijayanunni (from 1994 – 1999), even advocated for it. In the lead up to Census 2011, caste census advocates organised and gave a sustained push, to which the then Manmohan Singh-led government conceded to collect caste-wise data in the census. 

But technocratic reasoning by the Office of the Registrar General and Census Commissioner of India (ORGI) won out once again and the political leadership rerouted the caste-wise enumeration into a revamped Below Poverty Line (BPL) survey, which it renamed the 2011 Socio-Economic Caste Census (SEC survey). While the government published the BPL data in 2015, it deemed the caste-wise data from the  SEC survey unusable. More recently in the lead up to Census 2021, the ORGI announced that the census would enumerate OBCs, only to later walk back on its statement. 

The central government has once again announced it will include a caste-wise enumeration in the delayed Census 2021, now scheduled to occur in 2027. If the announcement translates into a change in census policy, it would indeed be historic. But institutional sabotage is still possible given the bureaucracy’s long-standing resistance to enumerating caste in censuses.

The official caste count in India increased from 3,208 in 1871 to 19,044 castes in the 1881 census. As of 2011 SECC, there are some 46 lakh castes. Why these variations? 

Across early censuses, the colonial state struggled to ‘commensurate’ caste—that is, create a common metric or set of comparable categories — within and across localised systems of caste hierarchy. Colonial census officials, for example, vacillated on whether to use the varna classification to collect and classify caste data across successive censuses. 

One practice that started during the colonial period was the development of regional caste lists prior to collecting census data from households. The creation of caste lists helped with commensuration while still allowing for the documentation of change over time. This practice continued in independent India with state-level SC and ST lists that are used during the decennial census. The 2011 Socio Economic Caste Census (SEC survey) – which was supposed to enumerate the caste of the entire population – only used SC and ST lists and made no effort to classify more than 75% of the population into a state-administrative category (e.g., ‘OBC’ or ‘general’) or create caste lists for these categories. When the data collectors recorded open-ended responses they were not instructed to standardise the spelling of caste names. Hence the enormous number of castes that you cite from the SEC survey.

A longstanding debate is over the right vehicle to collect caste data. The home ministry and Office of the Registrar General opined that mixing caste survey and the decennial census  will affect the “integrity of the census or distort the population count”. What explains this reasoning?

After leaders publicly agreed to collect caste data in Census 2011, census officials reframed the debate to focus on supposedly technical considerations. Their discussions over methodology sidelined the voices of ‘non experts’ and largely occurred outside the public eye. They used methodological concerns to convince the political leadership not to count caste in the census in order to protect “the integrity” of the population count, as you mention. Non-experts had difficulty challenging the premise that somehow enumerating caste in the census would create operational difficulties and ruin the integrity of the decadal population count. Technocratic reasoning (for which there is a high bar for entry) replaced a conversation about the importance of collecting caste-wise data – for the administration of affirmative action, broader policy purposes, and social justice concerns. 

I want to return for a moment to crucial texts I mentioned earlier, Jotirao Phule’s Slavery and BR Ambedkar’s States and Minorities, which describe the interwoven relationship between technocratic reasoning and Brahmanical power. Phule documented in detail the embeddedness of Brahmanical ways of thinking within the colonial state – how routine practices of oppressor castes humiliated and excluded caste oppressed groups from “public” resources and government services. Dr. Ambedkar argued that if the bureaucracy remained in the hands of those dominated by Brahmanical ways of thinking then social and political equality would not be possible in independent India. He foresaw how bureaucratic expertise and technocratic decision-making would reproduce existing hierarchies of power instead of dismantling the institutionalised mechanisms of exclusion. Up until the present day, we see examples of how this reasoning threatens the ability of citizens, particularly from historically marginalised groups, to have a voice in shaping the creation, design, implementation, and monitoring of public policies and programmes.

One of the recommendations put forth by the ORGI in September 2010 was to conduct a separate caste census. But this idea is practically and financially infeasible given the massive scale of human and financial resources already mobilised for the decennial census.

The census is the most experienced and competent project to collect all-India data on caste. The question is whether the ORGI is willing to do so – and collaborate with others who have a sociological understanding of caste and have been working to annihilate caste. Given the agency’s long-standing institutional resistance, it’s crucial that anti-caste scholar-activists provide oversight of the process and an alternative perspective to castelessness – an ideology that frames caste as a problem of oppressed groups and obscures caste privilege and power – which has structured recent institutional practices. This oversight will be necessary to develop an instrument and related protocols for counting privilege (for example, documenting well-to-do groups within the general category), along with enumerating groups that became eligible for central government reservations beginning in the 1990s (including caste-oppressed religious minorities such as Dalit Muslims and Dalit Christians), and groups that previous censuses have struggled to enumerate.  

You noted that we’re still collecting caste data – private groups do it, political parties map religious and caste identities to mobilise votes. What about a caste count as part of a decennial census would be different? 

While groups collect and use caste data for private purposes, collecting caste-wise data in the census would make the data available to the public and provide a crucial understanding of how caste intersects with gender, religion, education, employment, household amenities, place of residence, and family formation, among other socio-economic indicators in the census.  Caste-wise data from the decennial census could provide a more complete picture of how caste hierarchy structures socio-economic life in contemporary India. 

The census could document, for the first time since 1931, the relational nature of caste including how caste privilege and power operate. Caste-wise data on the “general” and OBC administrative categories, alongside the data on SC and ST categories, would challenge an ideology of castelessness that narrowly frames caste as a source of disadvantage and could inform policies and programmes to dismantle and destroy “caste apartheid”, a term caste abolitionist activist Thenmozhi Soundararajan uses to challenge the view that there are valuable and useful aspects of the caste system.

Since 1951, India has counted SC and ST but not other castes, and even within that there is no disaggregated data. Why did we pick just these two categories only?

Political leaders in independent India were eager to redirect time and energy away from the enumeration of caste, which they saw as an obsession of the colonial state but largely irrelevant in independent India. They argued that if the state focused on economic development, then caste hierarchy would dismantle itself; this view justified the decision to stop the caste-wise enumeration in the census. Dr. Ambedkar disagreed with the decision to exclude a caste count – his efforts to safeguard the interests of marginalised communities relied upon a detailed understanding of how caste inequality played out. 

The census has collected data on SCs and STs for the  administration of reservation policies. Following this logic, however, Census 2001 and 2011 should have also collected data on OBCs, given the administrative need for this data after the extension of central government reservations in the 1990s. That did not happen across two different political administrations. 

Activists argue the debate on caste census is narrowly framed between Dalits and non-Dalits and overlooks the contemporary caste landscape. Does this framing hide internal tensions in how different groups perceive and mobilise around the demand for caste enumeration?

The common framing of a caste census as an “OBC issue” largely ignores the need to document caste privilege and power. This remains true with regards to in the “general” category and among dominant groups within the OBC category.  

Documenting caste privilege is seen as optional and unnecessary, but data is also required on the general category given the extension of reservations to Economically Weaker Sections, which has been problematically limited to the general category (despite powerful arguments to the contrary). 

Your research found census data is not always objective: patriarchy and ‘castelessness’ of caste elites were embedded in the collection exercise. Can you give an example?

Castelessness was deeply embedded in the design of the 2011 SEC survey instrument. The survey consisted of one question on religion and two caste-related questions. The first question on caste included four answer-options (i.e., SC, ST, other, and no-caste/tribe) but excluded the state administrative categories of “OBC” and “general”. The “other” answer-option became a catch-all category including diverse groups from Dalit Muslims to caste-elite Brahmins. The first question failed to classify more than 75% of the population into a state-administrative category – a major shortcoming for a supposed “caste census.” 

The inclusion of the “other” and “no-caste/tribe” answer-option created an opt-out to the enumeration of caste. The SEC survey was designed to collect data from historically oppressed groups to determine their BPL status, as SC/ST status is one of the criteria by which the government ranks households not automatically included/excluded from BPL status. So the inclusion of the “no caste/tribe” option was to appease comparatively privileged groups who saw little benefit to having their caste identity enumerated. Comparative scholarship from other regions similarly shows that political and economic elites opt-out of documenting their wealth and resources at high rates. The answer-options “other” and “no-caste/tribe” in the first caste-related question prevented the expansion of data collection relevant for a caste-wise enumeration while continuing the longstanding practice of SC/ST data collection.

While the government has buried these data, we need to study them to understand the problems with the first nationwide caste count in independent India, and ensure the same mistakes are not made in Census 2027. 

Those against a caste census have argued that as a colonial instrument, it would promote ‘divide and rule’. How do you respond to this anxiety? 

Prominent anti-caste leaders have critiqued dominant streams of nationalism for more than 150 years – arguing that any quest for national unity that fails to explicitly tackle long-standing caste-based oppression or internal colonisation will reinforce Brahmanical ways of thinking, and in ways that often appear consistent with democratic values.

Censuses were tools of colonialism and extractive capitalism during the British rule, but they also helped to recognise political subjects who actively demanded that the colonial state address gross inequities and violence related to caste hierarchy. That’s why, for example, in 1955 Dr Ambedkar critiqued the absence of caste-wise data in the first census of independent India. As part of efforts to annihilate caste, he and others had used census data regularly to document the extent of caste-based inequalities and to develop anti-caste policies and laws. He criticised the home minister’s position that mistook the absence of census data as proof that caste was no longer significant social identity. At the same time, Dr Ambedkar also realised the census is a political instrument and knowledge-making institutions were controlled by dominant castes or those with “Brahmin spectacles”, a term Jotirao Phule used in the 1870s.

If we don’t have reliable caste data since the 1930s, how are allocations for welfare funds and quotas under different programmes being made?

After reservations were extended in the 1990s, policy makers and demographers have used projections from 1931 census data as well as data from recent sample surveys such as the National Sample Survey (NSS) to estimate the size of the OBC population, given the unavailability of the recent census data.

What can we learn from other countries (like South Africa) or other estimates that collect data on caste, race, and other social identities? 

Research from across the world shows that censuses are neither neutral counts of the population nor matters of bureaucratic routine. The entire process of census making is deeply political. Technologies play an important role in the production of census data, and while mistakenly presented as apolitical, they often reproduce dominant ways of thinking. Applying this to the Indian context, the 2011 SEC survey was touted for its technological advances including real-time electronic data entry (as data collection teams used tablets for the first time in a nation-wide survey) but produced caste-wise data that were deemed unusable for any public or policymaking purposes. Census 2027 is being called “the first digital census”, so we need to be wary of similar patterns unfolding, including where excitement over technology deflects time and attention away from necessary planning for a caste-wise enumeration.

The research also highlights how we classify caste-like systems, and for what purpose, is deeply embedded in historical understandings of citizenship and nation. Research on Latin America has found that census officials have played a key role in marshalling racial statistics to create and actualise a white-supremacist nation-building narrative of “progress”. At the same time, the absence of official racial classification can also be a powerful means of exclusion. This research pushes us to investigate whether in a particular historical moment a state’s system of classification for caste-like systems is designed and used to strengthen exclusionary projects or build inclusive communities. Applying these insights to the present day, some questions surface: is the current government willing to collect caste-wise data to document the extent of caste privilege and caste-based inequalities? Is it willing to implement and expand policies and programs to dismantle caste hierarchy? What type of institutional sabotage might occur and how else might the data be used?

How then do we conduct a successful caste census?

Given the ORGI’s resistance in the past, it’s particularly important that anti-caste scholar-activists serve on an oversight committee to advise and critically review the entire process. Starting with a detailed analysis of the religion and caste data from the 2011 SEC survey and the creation of castes lists for each state-administrative category in collaboration with regional governments and anti-caste organizations, and continuing throughout the design of the caste-related questions, their pretest, the development of interview protocols and enumerator trainings, and the collection, analysis, and compilation of data. The committee should also develop guidelines for ethical use of the caste-wise data and include both scholars who support the caste-wise data collection as well as anti-caste scholars who are opposed to a caste census. Ideally, when developing the instrument and protocols for processes of enumeration in the upcoming census, the knowledge and experiences of those most marginalised by caste hierarchy, patriarchy, and neoliberal capitalism will play a central role to help ensure the collected data will be used in support of policies and programmes to create more equitable societies.

With this type of an  oversight committee, the project will have a much greater  likelihood of navigating a range of challenges (including institutional sabotage) to document the extent of caste-related inequalities, and supporting the development and implementation of policies to annihilate caste.