(Info) Estimation of Poverty in Bundelkhand

Estimation of Poverty in Bundelkhand

Officially, poverty is measured in India in two ways and neither is very helpful in estimating poverty in Bundelkhand.

The Planning Commission defines poverty in terms of per capita monthly expenditure corresponding to per capita daily requirement of 2400 calories in rural areas and 2100 calories in urban areas. The per capita monthly expenditure is obtained from the National Sample Survey Organisation, which conducts an extensive consumer-expenditure National Sample Survey (NSS) every five years. A 'poverty line', originally calculated for 1973-74, is then revised, using a price index for rural and urban areas.

Leaving aside basic objections to this method of estimating poverty, as well as objections related to methodological issues  such as choice of price index, a limitation of this method is that it relies on sample surveys, which do not give us a good district-wise picture. We do get a region-wise picture, but  Bundelkhand, as we understand it (see Boundaries of Bundelkhand) does not constitute an NSS 'sample region'. All UP Bundelkhand districts do constitute an NSS sample region (in case of MP, the Bundelkhand districts are spread over three NSS sample regions) but the picture that emerges about UP Bundelkhand from NSS data is dubious.

According to NSS (50th Round) 1993-94 data, UP Bundelkhand was the poorest region of the state, with close to 70% of the population living below the poverty line. However, by NSS (55th Round) 1999-2000 data, the figure dropped drastically to around 27%. No world-record-breaking feat was achieved in the intervening years. The drastic drop was due to the fact that the two NSS surveys did not use the same design. Hence, straightforward comparisons between the two estimates are invalid. Conclusions derived from the two surveys, including one that suggests incidence of poverty in UP overall fell from 41% to 31% in this period, are not taken seriously.

Some methods have been suggested for 'correcting' the 1999-2000 estimates to make them comparable with the 1993-94 estimates. The 'corrected' rate of incidence of poverty in rural and urban UP Bundelkhand, quoted in the 2003 Human Development Report- Uttar Pradesh, is around 38%. The correction however does not take into account a basic problem: to start with, the number of sample households from Bundelkhand was very small. Hence, as a 2002 World Bank study on poverty in UP noted, 'sampling errors are likely to be large' [World Bank, p 20 footnote].

Another problem, not specific to Bundelkhand but relevant to the region, is that in 1993-94, there was a high concentration of households around the poverty line. A little economic growth could push a large number of people above the poverty line. However, many of those who 'got pushed up' from below the poverty level continue to be vulnerable to 'shocks' that can pull them back to  below the poverty line [World Bank, ii]. And Bundelkhand is highly prone to weather-induced shocks (see Impact of Bundelkhand's Geo Profile on Human Life).

BPL Survey

The other official method of estimating poverty is also contentious. The Union ministry of rural development has been estimating rural households living below the poverty line (BPL) every five years since 1992. While the NSS estimates percentage of poor, the BPL survey seeks to identify particular BPL households in rural areas, which then become eligible for various social welfare scheme benefits, including subsidised foodgrains through the public distribution system (PDS).

Following criticism of the design of the 1992 and 1997 BPL surveys, the Union government appointed an expert group, which suggested use of a schedule with 13 socio-economic indicators, to be filled in through house-to-house surveys. The indicators related to factors like size of land holding, type of house, availability of clothing per person, food security, sanitation, possession of consumer durables, literacy, status of household in labour force and means of livelihood. Against each indicator, a household was given a score of  0 to 4. The lowest aggregate score thus was 0; the highest was 52. There was also a 'cut-off' score to determine whether a family falls below or above the poverty line.

This method, used in 2002, had many problems. First, there were design issues, in the scoring system. If a family ate less than one square meal a day, it got 0 on this parameter, but a family that ate twice a day got a high score of 3, even if it faced occasional shortages. Similarly, a family with a one adult male and female earning-member got a high score, regardless of the income earned. Implicit in the scoring system was the assumption that all indicators carry equal weight. Hence, having less than one meal a day was given the same weight as not owning specified consumer durables like electrical appliances - which is quite absurd.

There were also problems in the way the questions were framed, which would have been reflected in data collected. There are fine distinctions between having 'less than one square meal per day for major part of the year' and 'normally, one square meal per day, but less than one square meal occasionally', which are likely to be missed by respondents if the surveyor is in a hurry - which is more than likely. Most questions were highly specific, but in some cases, there was considerable scope for interpretation. What, for example, is a 'major part of the year'? Six months, or nine?

A major problem related to cut-off scores. There was no common cut-off score across the country. State governments were allowed to specify cut-off scores, and it could vary from village to village. Predictably, there were serious allegations of different cut-off scores being used for political reasons. Cut-off scores were also determined by administrative and fiscal reasons- by the number of BPL households to which a state government could provide due benefits.

There was an overriding cut-off predetermined by the Planning Commission, on the basis of  the NSS data: in no case could percentage of BPL households in a state exceed the poverty ratio of the state determined according to NSS data. (The sheer absurdity of mixing up two surveys that used different methods, for different purposes, was overlooked). Accordingly, in many states, district administrations were given 'limits' to maximum number of BPL households that could be identified in particular blocks.

In effect, the 2002 BPL Survey did not attempt to identify BPL households. It only attempted to identify households that would be eligible for BPL benefits, according to the state government's capacity and interest. And many of these households were wrongly identified, deliberately, to please vested interests and get election benefits.

For good reason, the Supreme Court ordered a stay on the use of the survey, following a writ petition filed by the People's Union for Civil Liberties (PUCL). The stay was lifted in February 2006, and various half measures were suggested to resolve problems related to the survey. While the 2002 survey results would form the basis for benefits under centre government schemes, state governments were free to adopt any other criteria/survey for choosing beneficiaries for their schemes. Complaints were invited and re-surveys were ordered.

In MP, the government decided that the BPL list would be prepared in such a way that it would include all households that were in the 1997-98 list - no matter that a survey new design was used in 2002 as the earlier design was found to be flawed. In UP,  a completely new survey was ordered by Mayawati after she came to power, but the survey was suspended by her in October 2008, after the opposition alleged that landless and homeless households among the upper castes and OBCs were being deleted from the BPL list, and households supporting the ruling party were being included indiscriminately.

To cut short a long story that has seemingly no end, we have no reliable and acceptable way of estimating poverty in Bundelkhand. For whatever it is worth, the poverty level in Bundelkhand districts, on the basis of  pre-determined cutoff 2002 BPL survey scores, ranged from around 21% in Mahoba to 55% in Chitrakoot, in UP and 17% in Datia to 52% in Damoh, in MP. The figures for Mahoba and Datia are absurd.

However, responses to particular questions, discussed in 2002 BPL Survey Data, do give us a fairly good picture of  poverty across the region, if we look only at the data, and not the overall scores, and treat the data cautiously.

Go Back To Main Page

Courtesy : bundelkhandinfo.org