Tuesday, May 11, 2010

Will counting caste help to reduce inequality?

More thought needs to be given to the kind of data generated and its practical implications

Yesterday when the census enumerator visited, I asked him how he felt about the current debate on counting caste in the census: “Not comfortable at all”, he said, “I don’t even like asking whether someone is SC/ST or Other, leave alone what their caste is”. But, he added, “caste is an inescapable reality of Indian society.”

The debate on counting caste in the census has not moved on from 2001, when opinion was equally divided. Supporters of caste enumeration argue that census categories merely reflect existing classifications, and that only the census can provide the figures necessary to map inequality by caste. Opponents argue that the census does not mirror but actively produces social classifications and ways of thinking.  They point to the history of mobilisation around caste in the census and the consequent dangers of both distorted data and increased social tensions. In neither case has much thought been given to how the data might be used, the different kinds of figures needed for different purposes, or alternative ways of collecting the required data.

On the surface, caste enumeration appears to be a UPA concession to its OBC allies, but more fundamentally, it fits with the larger political agenda of moving people off the land, holding out the illusory promise of formal employment. For social justice, we are made to believe there is no alternative to reservation, and for reservation, no alternative to counting caste. With over 90% of people in the informal sector, reservation can hardly be the primary solution to greater equality. There is no doubt that stringent affirmative action policies are required to make formal institutions more socially inclusive, but to shackle the census to this agenda betrays a failure to learn from the past or to think imaginatively about the future.

University degrees are important for certification, especially for those historically deprived of education, but they do not necessarily contribute to the creation and expansion of knowledge. For instance, there are over 20,000 rice varieties in Chhattisgarh, some 6000 of them in Bastar alone, yet this knowledge is rarely factored into discussions around educational expansion. ‘Social Justice’ becomes simply whether certain castes get admission into agricultural universities, not whether those insitutions enhance existing knowledge or contribute to people’s well being. And in the meantime, the holders of such knowledge are being decimated through land acquisition, displacement and inhumane forms of counterinsurgency. The counting of SCs and STs in the census has not led to any greater justice for them -  not only do Mirchpur type incidents continue; but even in terms of planning or the everyday provision of services in villages, common educational or health facilties are often situated in upper caste hamlets, even when there are clearly larger populations of Muslims or Dalits in the village.

The transformation of caste through the census

While earlier rulers also created lists of castes and occupations, such as those in the Ain-i Akbari or the Rajatarangini, the urge to map every single caste is commonly attributed to the colonial need to know their populations in order to govern. Caste and religion were seen as key categories with which to explain native behaviour: to explain insanity, to help in the recruitment of ‘martial races’ to the army, or to determine which groups had a propensity to crime.

Yet successive Census Commissioners like Risley in 1901 and Yeatts in 1941 described the caste tables as the most troublesome and expensive part of the census: Risley complained: "If the person enumerated gives the name of a well known tribe or caste...all is well. But he.. may give the name of a sect, of a sub-caste, of an exogamous sept or section....his occupation or the province from which he comes.” In 1881 in Madras presidency alone, the inhabitants returned 3208 different castes, which the census then regrouped into 309 castes.

Enumeration also required people to be slotted into categories that were mutually exclusive even if untrue to their lived experience. A person could not have two castes or two religions. Where the 1911 census had recognised several sects as Hindu-Muhammadans, in 1921 they were reclassified as either one or the other, except for the Sindh Sanjogis who refused and were relegated to `other'. The Meos today face similar problems, caught between the Vishwa Hindu Parishad and the Tabligh e Jamaat.

As people began to realise the value of census categories for economic, social and political advancement, mobilisation around the census increased, particularly after Risley’s 1901 ranking of castes in order of “native opinion of social precedence”. Numerous petitions to the census commissioners asked to have the names of castes changed or be ranked higher in the social hierarchy. For instance, the Khatris claimed that their name was really a corruption of Kshatriya. The census also initiated a wider transformation, with hundreds of caste associations formed between the 1880s and 1930s, addressing their demands both to the state and towards internal social reform.

Caste was not the only ascriptive identity politicised by the census. Religion, especially pre-partition, and language were equally explosive, and saw complaints against alleged enumerator bias. For instance, in 1941, the Dalit Chuhras in Punjab complained of pressure to be recorded as Sikhs or Hindus by Sikh and Hindu enumerators and demanded that their religion be entered as Adh Dharm instead.

Given such battles, and the concern that India’s innumerable castes and religions were used to justify colonial rule, the constituent assembly framing the Census Act of 1948 decided to exclude caste returns (except for SCs and STs). However, since caste did not disappear from public life as was hoped, political attitudes towards counting it have changed dramatically. Similar debates have taken place over the counting of race and ethnicity in the US and UK census respectively, with some people pointing to the unscientific nature of race, and others arguing that "In order to get beyond racism, we must first take a count of race."

Nature of Data

Assuming (optimistically) that the demand for caste enumeration is driven by an anti-discrimination, pro-equality impulse, we need to consider how the data thrown up by the census will fit public needs. Unlike earlier censuses which were caste and religion based, any proposed caste inclusive census would not have caste as a key variable, but simply as one return among others. It will depend on the precise tabulations decided upon whether we get a caste wise breakup of literacy, sex ratio, female work force participation etc. One argument being made is that it will help to identify weaker castes among the OBCs, but that would depend on the level of caste detail (sub-caste, caste) at which tabulations are carried out.

The major benefit the census will provide is the numbers of each caste by region, making it possible for researchers to conduct other kinds of surveys, e.g. to assess through additional sample surveys, the percentage of civil servants from a particular caste. On the other hand, because of returns which fluctuate according to identity politics, it may be difficult to construct accurate time series records  to assess changing mobility trends.

Even for the purpose of measuring ‘backwardness’, the census is only a beginning, not an end in itself. While the Mandal Commission extrapolated from 1891 and 1931 census data, this alone was not the basis for its classifications. The comprehensive socio-economic survey conducted by the Second Backward Classes Commission (BCC) in Karnataka under Justice Venkataswamy yielded generally accepted population figures for each caste, but its indicators of backwardness were flawed. As Justice Chinappa Reddy, who chaired the Third BCC noted, simply aggregating all the indicators of backwardness (data potentially available through a census) and ranking castes on that basis as was done by the 2nd BCC would place Vokkaligas in Karnataka on par with Darzis. The Third BCC therefore developed its own indicators of backwardness on the basis of several different kinds of data which included: personal touring; representations from caste associations; a sample socio-economic survey covering 600 villages; information from taluks on caste wise land holding; survey of caste and socio-economic background of gazetted officers, MPs, MLAs, leading Advocates, Professors, etc.; information on caste, occupation and income of parents for students appearing in the SSLC exam; information on admissions into medical, engineering, dental colleges, etc; and information from the Karnataka Public service commission and other recruitment agencies on 3.47 lakh government employees and 1.20 lakh public sector employees.

In short, while the census can provide base figures, it cannot substitute for the kind of information needed both for inclusion of castes in an OBC list or for `graduation’ of castes out of the list, even assuming the latter were ever to be politically feasible. In his discussion of sources, Justice Chinappa Reddy pondered over the wisdom of excluding caste from the census, noting that such data would have saved the commission many problems. However, he went on to add: "On closer thought, I think it is just as well that caste is ignored in the census operations. A beginning has to be made somewhere to forget caste."

The author is Professor of Sociology at the Delhi School of Economics.