Thursday, May 27, 2010

Ways to accurately estimate the OBCs in India

There is a renewed demand from many corners, especially a segment of political leadership to collect a citizen’s caste affiliation during the 2011 census of India. The GOI in principle has agreed to collect information on caste in 2011 census; but what is not clear is as to what kind of information on caste will be collected; will it be collected during the house listing operation or during the population census.
Operational Difficulties in Collecting Caste in Census 2011: The census is undertaken in two phases. A house listing operation (Phase 1) precedes the population enumeration (phase 2) usually scheduled during months of February and March of the census years. The house listing for census 2011 is already in progress. House Listing Operation collects information on SCs & STs - item 15 in the proforma. Once can expand this question by providing an addition option ‘OBC (3) and changing code of ‘Other’ to (4). Note also that information on Scheduled Caste is collected only from Hindus, Sikhs and Buddhists and not from other religion; whereas Schedule Tribe can be from any religion. Since religion is not collected in this phase the OBC reporting will be free of divergent influences and overall this approach can work if the OBC check list is provided to enumerators in advance. But since the listing operation is already on all over India and it is too to collect caste/OBC information during the house listing operation.

According to news paper reports, it appears, that data on caste will be collected by asking a simple question “what is your cast?” during the census operations. This now can be done only in phase 2 known as Population Enumeration undertaken during March-April 2011. This kind of open question can be asked immediacy after the following questions:
Q. 7: Religion (Write name of the religion in full)
Q. 8: If Scheduled Caste, Write name of the SC from the list supplied
Q .9: If Scheduled Tribe, write name of the ST from the list supplied
Proposed Q.10: What is your Caste? _____________________

Such a question will generate a lot of caste reporting exclusive to each state and regions within state. Often a given caste is identified, pronounced and spelled differently; and sometimes same cast can assume differential status in other states. Collating such information subsequent to conclusion of Census will be a herculean effort and also cause for controversies. Therefore one needs to be clear and careful as to what kind of caste data to be collected during the population enumeration.

Therefore it is suggested that the Proposed Q.10 during enumeration phase should collect information only on ‘as to whether the individual belongs to a OBC category? This is best done if an approved list of OBCs is provided in advance. In the absence of approved OBC list, self-reported post enumeration categorization will be controversial. Yet eliciting OBC information during enumeration will have to be done as an additional question after the question on religion there can be errors due to misreporting and confounding effects. Recent debates on extension of ‘reservation’ benefits to individuals belonging to religions other than Hindus have generated anxiety and also expectations especially amongst the minorities.

Possible Contamination of Census Information : India has a long uninterrupted history of undertaking census every 10 years since the late 19th century and certainly a regular pan-Indian census since the Independence. Indian Census is respected across the world for its quality and academic content. A lot of debates and research has fed the number and type of question to be included in the ‘house listing’ and ‘population enumeration’ exercises. It is useful to note that the last time caste information was collected was during census 1931.

The concepts, definitions and questions are standardized in such a way that the inter-censal comparisons are possible. Such comparisons are the basis to evaluate a number of economic, social, educational and work/employment parameters for India and its many states and even districts. Therefore, any addition or alteration and even change in placement of a question, both in the listing and enumeration proforme can cause changes in the quality of data in other words can contaminate census information. For example, (a) the errors in self-reported SCs/STs and OBCs category can be enormous by boosting the respective population shares which can prompt increase in the quota shares. This can happen because – those not SCs/STs may get motivated to report themselves as such; so will be the case for OBCs. Since the OBC reservation debate is in its peak one can expect an extra caution amongst the OBCs to ensure reporting; while there can also be misreporting thus boosting the share of OBCs in the population. Even communities with no caste identities may innovate or identify a caste for themselves. (b) It is likely that caste question can affect the reporting of occupations. The possibility is large due to expected gains offered by the government based on caste/occupation linked targeting such as programs for ‘weavers’ and so on. There can many other distortions or contamination of data.

What can be Done? It appears the demand for caste census emerges from the need to know the accurate size of the OBC communities. Since data on the SCs and STs are regularly collected since 1951 census, there is no issue relating to these categories. Thus the focus should be on identifying correct size and share of OBCs, since the Mandal commission linked reservations are based on old data from 1931 and not so valid local level surveys. The quinnqunnel National Sample Surveys are undertaken about five years apart and they are most respected for quality and consistency of data. Current sample size is over one hundred and fifty thousand households rural and urban households. Fortunately, India also hosts independent survey research organizations such as the National Council of Applied Economic Research which can execute such surveys with ease and accuracy. Should there is a need one can devise a multi-level multi-stratum sample sizes so as to provide the estimates of the religion specific size and share of OBCs in each of the Indian state and regions within the larger states of India. Such estimates will keep the chastity of the Indian census intact while proving accurate data for devising pro-poor social policies in India.

No comments: