Why India needs caste-based census

Even countries with good recording systems for births and deaths conduct a population census once in ten years. These snapshot counts capture population movements in a more elegant way than having a police state that monitors movements within a country. And several countries go beyond the minimalist census to record information about all manner of things, including household income.

In developing countries, the census is the only source of data on demographics, and any other information needed to inform public policy. The 2011 Indian census, for the first time since Independence, is considering collection of caste data. This has generated some heat, as might be expected.

Had there not been a resurgence of caste-based quotas after 2006, there would have been no need whatever for introducing it into the census. But after 2006, accurate knowledge of caste shares in the total population is critical for equitable fixation of quotas. Even countries practising merely equal opportunity, with no preferential affirmative action, let alone quotas which are the last step up the positive discrimination ladder, conduct a baseline count along the relevant boundaries. The United States census, currently under way, routinely includes a question on race. The information is collected as self-declared, and is treated purely as an identity issue entirely independent of visual markers, or fractional ancestry.

No quota can operate in a data vacuum.  Indeed, constitutional equality is violated if a caste-based quota is mandated in the absence of a uniform nationwide database, of the kind only a census can provide. Caste enumeration in the 2011 census is essentially dictated by the quota decisions taken earlier. No country can put in place job and education quotas for an underprivileged segment of the population, and then bury its head in the sand about the boundaries along which those entitlements are drawn.

Among the urban elite, caste is seen as private information, of the kind people may choose to reveal in certain contexts like matrimonial advertisements, but which they should not be compelled to reveal to a census enumerator. This can be easily accommodated. There will clearly be a box for unknown or undeclared caste. Indeed, this freedom should be extended to questions on religion and sex. A third box for the last, in particular, would establish that the sum of males and females need not equal the total population, and so recognise the rights of the mocked small minority in every country that is of indeterminate gender.

The majority of residents know their caste identity more clearly and unambiguously than they know their own age. I have marvelled at the ability of census enumerators to ferret out the age of individuals, by piecing together answers, often mutually conflicting, related to religious cycles like the Kumbh, or by probing personal memories of major events. Between age and caste, age is unquestionably the more intrusive query. Even names in India are ambiguous, with public and private variants, and with the first names of women radically altered after marriage among certain groups, in accordance with custom.

Caste by contrast is uniquely assigned, and is public information in any rural settlement. Although caste consciousness is decidedly lower in urban India, caste identity of households is publicly known in all but a small minority of urban residential neighbourhoods as well. Clearly, caste is neither relevant, nor sought to be known in anonymous urban spaces like worksites or commercial areas. But it is known in places of residence, which is where the census is conducted.

The stand that caste consciousness, with its associative aspects of exclusion and entitlement, is best obliterated over time by neglect would have been tenable but for the resurgence of quota consciousness. People are already familiar with classification into categories like scheduled, backward and general, in contexts like admissions or employment registration. The new question will just be a more factual version of information they are accustomed to providing.

Aside from yielding the database for defining caste-based quotas, knowing caste shares in the population will enable a more fair operation of the reservation among the beneficiaries, who are a mutually unconnected set of castes. The same network exclusion effects which led to the introduction of quotas in the first place will clearly operate within the quota. Scheduled and backward reservation can be dominated by a few well-networked groups. The purpose of the quota will only be served if there is a database on representation of the stipulated groups in the population, and their representation in the quotas designed to give them privileged access.

The best analogy is the Generalised System of Preferences), a trade preference scheme operated under what was formerly GATT (now the World Trade Organisation), whereby importing countries in the developed world were permitted to offer positive discrimination towards developing country exporters in the form of reduced tariffs, usually set at zero. Because the general MFN tariff extended without discrimination had been lowered through successive rounds of trade negotiation, the differential advantage of a zero tariff GSP came down to somewhere between 1 and 4 per cent, depending on the product.  Only a few developing countries were close enough to compete effectively at such a narrow price advantage, so that GSP imports came to be dominated by them. A graduation provision was, therefore, introduced, usually set at half of total imports under the GSP. Even where Indian exporters were among those graduated out, they had to concede the fairness of the provision.

Quotas are designed to give forcible entry for underprivileged groups in a system which poses multiple obstacles to their advancement. If access to the quota is uneven across those groups for whom it is intended, clearly it only replicates the injustices of the larger system. But unevenness of access itself is impossible to assess without a census of population shares of beneficiary groups. If caste information is not collected, we will have denied ourselves an opportunity to make the quota function in an equitable manner.

The author is honorary visiting professor, Indian Statistical Institute, Delhi