The Cabinet cleared the proposal to have a new BPL Census to see how many families in the country are poor. The Census is to be combined with one on caste, to ascertain the asset-ownership of various caste groups. We examine the implications of the move

Why do we need another estimate of the number of poor?

Right now, there are various estimates of the number of poor. While the Planning Commission estimates this at 27.5% in 2004-05, Professor Tendulkar put this at 37.2% and the late Dr Arjun Sengupta put the poverty estimate at 77% of the population.

How do we get such different estimates, and how will the BPL Census help?

Right now, the poverty estimates are based on the National Sample Survey?s (NSS) data of what Indians in different groups consume. This large survey is done every five years. While the Planning Commission estimate was based on the current expenditure levels below which people are considered poor (this is R356.3 in rural areas and R538.6 in urban areas), Professor Tendulkar?s committee felt the urban estimates were correct while the rural ones were not?it then made a correction for this. Others, like Surjit Bhalla, argue the NSS data underestimates the amount of consumption in the economy by over half, and so you need to correct for this. Once you do this, the poverty estimate drops dramatically.

A Census, by definition, gets over the problem since it counts each and every household in the country and gets asset details from it.

So why haven?t we done a BPL Census before?

We have, in 1992, 1997 and 2002.

And what happened?

In each case, experts rejected the results on one ground or the other. So the results didn?t really help.

The first BPL survey for rural areas, in 1992, used income as the main criteria and all households with incomes of less than R11,000 were classified as poor. The number of poor estimated by the survey was almost twice the number of BPL population identified by the Planning Commission. But even then, the family-based fixed poverty line created problems for large families resulting in the exclusion of many poor families.

How was this fixed the next time around?

The 1997 BPL Census changed the identification criterion from income to consumption and from a household to a person basis. A set of exclusion criteria, like ownership of two or more hectares of land, pucca houses or ownership of TV, refrigerator, two or three wheelers and farm equipment, was also further used to summarily eliminate ineligible families from the list. Families whose consumption was within the norms set by the Planning Commission were classified as BPL.

The third BPL survey, in 2002, used a score-based ranking using both social and economic indicators for identification of poor households. As many as 13 parameters were used to assess the quality of life of households, with scores ranging from zero to four for each parameter; the absolute and the relative position of each household in a village was set on the basis of the aggregate score, which varied across villages, blocks and districts. State governments were also given the option of deciding the total number of poor households in conformity with the 1999-2000 figures on poverty or the adjusted share computed by the Planning Commission. In addition, the states were also given the flexibility of adding another 10% to account for the transient poor.

What do the experts have to say about all that?s been done so far?

An expert group set up to advise the government on the new BPL Census argued that the 2002 Census parameters were incorrect. The one on landholdings, for instance, did not distinguish between private irrigation and public irrigation; nor did it look at the quality of land while deciding whether a family was poor; the possession of consumer durables was incorrectly applied?possession of a colour TV, the group said, required a lot more purchasing power than a radio, even though both were consumer durables. Based on this, and a comparison of the BPL Census data with that of the NSS, the group concluded ?61% of households?who are poor on the basis of their consumption expenditure being less than the official poverty line?are excluded from the net of BPL Census?.

Even more damming was what the group said about the quality of the data: ?And lastly, the requirement for staff for doing a comprehensive survey was not worked out with the result that often in actual practice no detailed survey was done, and survey sheets were filled up within the office itself.?

Yoshifumi Usami, Biplab Sarkar and VK Ramachandran did another interesting comparison of the BPL Census data for 2002 in four villages in Maharashtra, Uttar Pradesh, Rajasthan and Andhra Pradesh. The data was compared with that collected by the researchers and they found in some cases, the number of BPL households in a village was greater than the number of households in the village by over a third; in other cases, they found the number of poor households was overstated by as much as 45%.

So will things be better in the 2011 BPL Census?

Let?s hope so! Certainly the government is taking better care. For one, the Census?s listing of households will be used for the BPL Census. Two, the expert group has made many suggestions. The proof of the pudding, however, lies in the eating. Fortunately, we will get the results of the Census done by the Census of India at around the same time we get the results of the BPL Census. So some kind of comparisons can be made.

What about the data on caste?

Adding the caste dimension makes the BPL Census more tricky. Once the BPL/Caste Census is done, it will give rise to reports like, say, 11.38% of SC households are poor, and just 3.97% have managerial jobs. This will raise even more demands for caste-based affirmative action. Apart from the methodological issues that come up in terms of how valid the BPL/Caste Census data will be, since it is not being done by the Census of India, the data needs to be interpreted carefully. Merely reporting that 3.97%, or even 79.3% of SC households have managerial jobs is irrelevant unless it is linked to the proportion that have higher education degrees; the fact that more SC are poor than OBCs may have less to do with caste than with the fact that more SCs live in villages and are less educated. Even Census data, as we?ve seen in the case of previous BPL Censuses have to be interpreted with care.