Poverty’s definitional woes

Poverty’s definitional woes

Poverty estimates stumble on differing definitions of the household for statistical and policy purposes

Himanshu

Officially, the Planning Commission accepted the Tendulkar committee’s report on revision of poverty estimates after the empowered group of ministers on food security asked the commission to issue a final estimate of poverty in the country. Despite the commission’s acceptance, the ministerial group asked it for another estimate of poor households.

The simple reason is that the Tendulkar committee’s estimate— or any estimate by the Planning Commission—shows the percentage of poor people in each state. The number of poor households has never been released.

In theory, once the percentage of poor people is known, it should be straightforward to arrive at the number of poor households. In reality, the process is complicated by the fact that households are defined differently for different purposes. While the National Sample Survey Organisation and the census recognize a common kitchen as the basis of identifying a household, the below-poverty-line (BPL) census and the Mahatma Gandhi National Rural Employment Guarantee Scheme (MGNREGS) use the definition of a nuclear household. The extent of variation due to this difference is huge.

Take, for example, Bihar— among the poorest states in the country. Food ministry data says Bihar has 6.42 million BPL/AAY (Antyodaya Anna Yojana) cardholders. Since the existing number of BPL/AAY beneficiaries is fixed, according to 1993-94 poverty estimates, this amounts to 54.96% of households in Bihar. This percentage— actually a percentage of the poor population—is then converted to the number of households using the estimated household number in Bihar as 11.9 million in 2000. This estimate is arrived at using the census population estimate of Bihar at 73.1 million and an average household size of 6.16. That means the Bihar government has almost exhausted the number of beneficiary households that the BPL census identified correctly or incorrectly.

Of course, there were many targeting errors in this process. When the Bihar government undertook the BPL census (after a Supreme Court stay had been lifted in 2006) it found that the total number of households to have been surveyed was 22 million, with the number of beneficiary households at 13 million. As a result, the state government had to exclude almost half of the poor correctly identified. This was almost double the existing number of households that the food and rural development ministries recognize as poor. The average household size reported by the state’s BPL census turned out to be 3.99—much lower than the 6.16 estimate used by the Indian government.

The Bihar government has pointed this out and its chief minister has been critical of the Planning Commission’s poverty estimate. However, the problem lay with the estimate of household size, which in turn led to underestimation in the number of households in the state. Had the number of households in the state been correctly recognized as 22 million based on the average size of 3.99 and adjusted population estimate, all the beneficiaries identified by the state government could have been given their entitlement. That is, the state government should have insisted on correcting the estimate of household size and total number of households in the state.

Now that the poverty estimates have been corrected, does it make a difference? In the case of Bihar, it unfortunately does not. The new poverty estimate based on the Tendulkar committee report for the state is 54.5% of the population— roughly the same as the the old poverty estimate of 1993-94.

The real problem lies in the different ways households are defined for statistical and policy purposes. This is something that has been imposed by the programme design, be it MGNREGS or the public distribution system, where entitlements are based on households irrespective of their size. It, therefore, makes sense for households to show themselves as two different units even though they may be sharing the kitchen.

This would not have been a major problem if it were just a convenience of nomenclature and definitions. But this has an impact on actual services obtained, particularly for the poor, who are forced to remain outside the purview of programme benefits because of arbitrary caps assuming statistical estimates of household size. But this also has the adverse social consequence of young married people not wanting to live with their parents because it deprives them of government benefits, or siblings being forced to split households to avail of social benefits. It will then not be unfair to say that our programme designs have contributed to splitting households for pecuniary benefits. While nuclearization has often been a subject of enquiry for sociologists, it would be interesting to analyse the impact of public programmes-driven division of households.

There is also a message here for the policymakers: It would be much better to design basic entitlements such as employment and food on an individual rather than household basis. For the proposed right to food programme, this would not only take care of the unnecessary debate on the 25kg versus 35kg entitlement, it will also mean that food subsidy is utilized properly. Even more importantly, it will be in the true spirit of a right that is an individual and not a household attribute.

Comments are welcome at theirview@livemint.com

Himanshu is an assistant professor at Jawaharlal Nehru University and a visiting fellow at Centre de Sciences Humaines, New Delhi.