Amidst the din about corruption, inflation and shooting by Chhota Rajan, an important piece of news received relatively less attention- launch of the fourth national socio-economic census, more popularly known as the BPL Census on June 29 in a non-descript block Hezamara in Tripura.

The objective of the exercise is to identify ‘poor’ households, who can be provided assistance under various programmes of the Central and the State governments. This round of Census is significant since

a)      the methodology of identifying ‘poor’ has been modified and improved compared to the previous BPL Census,

b)     for the first time, the Census will cover urban areas,

c)      for the first time, information about the caste of the household would also be obtained.

While it may seem easy to spot poor households in a village, the matter is not as simple when it comes to actually identifying them as ‘poor’. The methodology followed has evolved significantly since the days of the first BPL Census conducted in 1992.

BPL Census 1992

This was a relatively simple affair, which used annual income of the household as the criterion for classifying households. A household with a total income of less than Rs. 11,000 per year was considered poor. The Census estimated that 52.49% of the rural households were poor by this yardstick, almost twice of the ratio estimated by the Planning Commission.

Of course there were several issues-

a)      A family-based fixed poverty line created problems for large families with small income per head resulting in possible exclusion of poor families from the BPL category (and the opposite for the smaller families),

b)     the methodology ignored the multi-dimensional nature of poverty,

c)      given the benefits associated with the ‘BPL’ status, the respondents would have incentive to underreport his/her income, leading to over-estimation of poverty.

Realising these issues, an expert group was set up whose recommendations were incorporated in the BPL Census that was carried out in 1997.

BPL Census 1997

A number of improvements were made in this round – the criterion of determining the cut-off was changed from income to consumption; the poverty line basis was changed from household to per capita. Most importantly, before administering the questionnaire to a household to determine its consumption expenditure, a set of exclusion criteria was applied to eliminate ‘non-poor’ families straightaway. Thus, if a household was found to possess certain assets, it would be regarded as a ‘non-poor’ household and excluded from further investigations. Information about total consumption expenditure was then obtained from the remaining households. The per capita consumption expenditure so derived was compared with the (state) poverty line as derived by the Planning Commission). If the former was less than the latter, the household would be identified as a ‘BPL’ household.

This Census also estimated that 41.05% of households were poor as against 26.10% estimated by the NSSO.

Though improved, this Census had its share of criticism. The strongest criticism was against the exclusion criteria, which was regarded too stringent- possession of a ceiling fan would lead to exclusion of the household. There was no provision for adding to the BPL list by way of inclusion of households who might become poor subsequently. The exclusion criteria did not allow any geographical variations. For example, one of the exclusion criteria was whether the household operated more than 2 hectares of land. But it did not take into account the fact that income generating potential of an irrigated and non-irrigated land of the same size is vastly different.

To address these criticisms, another expert group was formed in 2002 before the next round of the Census implemented in 2006.

BPL Census 2006

The Census dropped the exclusion criteria. Instead, a score-based ranking was adopted where each household was to be ranked based on 13 indicators. They are –

1. size of landholding,
2. type of house,
3. clothing,
4. food availability,
5. sanitation,
6. ownership of consumer durables,
7. literacy status,
8. status of labour,
9. Means of livelihood
10. Status of Children
11. loans from informal sources for daily consumption,
12. migration and finally,
13. preference for assistance.

The response to each question was scored on a scale of 0 to 4 and then aggregated. Thus, a household could score anywhere from zero to 52. The cut-off scores to identify poor were allowed to vary from state to state, subject to the constraint that ‘total number of poor persons identified in a state/ UT does not exceed the number of persons living below the poverty line in that state/ UT by more than 10% of the estimates prepared by the Planning Commission.

This methodology was severely criticised-

i)                   Even this Census did not allow for addition to the BPL list, while deletions were allowed.

ii)                 Dropping the exclusion criteria altogether increased costs and time required for enumeration. If the previous exclusion criteria were regarded as stringent, then as pointed out by some of the researchers, the solution was to modify it rather than drop it altogether.

iii)               The most serious problem was simple aggregation of scores which ignored the notion of hierarchy of basic needs. The aggregation established ‘cardinal equivalences across ordinal rankings of alternative states of households in respect of individual indicators’. Thus, having ‘less than one square meal a day for major part of the year’ gets equated with ‘open defacation’.

iv)               Questions were raised about the very utility of such an exercise. Information on a number of indicators was already available through the 2001 Census. This information could have been used for social programmes which didn’t necessarily require household level information. Even for the schemes where ranking of households would have mattered (for schemes such as ‘Antyodaya’), aggregate scores completely masked the relevant information (in this case, food availability for the household). Further, a number of anti-poverty programmes are self-targeting in nature and ranking of the households doesn’t matter.

Given this context, it would be interesting to see the methodology adopted for the latest BPL Census, the one which started on June 29.