The Great Indian Poverty Debate: Fresh estimates ignite new discussions

A recent State Bank of India (SBI) study says that India’s poverty was well below 5 per cent in 2023-24. This was after the government released a fact sheet from the 2023-24 Household Consumer Expenditure Survey (HCES), recording a decline in poverty from 7.2 per cent in 2022-23 to 4.9 per cent in rural areas, and from 4.6 per cent to 4.1 per cent in urban areas during the same period. 

The decline in poverty at the national level, by 17.2 percentage points, however, would be considered very sharp during the period from 2011-12 to 2023-24, with the poverty figure in the base year being 21.5 per cent. 

In the past, the poverty debate in India had revolved around the release of the HCES reports roughly every five years. These surveys had a reasonably uniform data collection methodology from the 1970s to the mid-1990s. There was a general acceptance of the concept of a poverty line corresponding to a minimum calorie intake norm following the initial work done by Vinayak Mahadev Dandekar and Nilkanth Rath in the 1960s. Subsequently, the then Planning Commission adopted the methodology with some modifications for the official poverty estimates. 

However, the widening mismatch between the household consumption expenditure in the National Sample Survey (NSS) and that in the national accounts statistics forced a review of the survey methodology, specifically the reference period used for recording the expenditure. This and the subsequent attempts to review the methodology of defining the poverty line through the recommendations of the Suresh Tendulkar and C Rangarajan committees somewhat unsettled the entire debate. This has made the recent poverty debates a complex exercise of juggling data and methodology, leading to contradictory conclusions. 

The latest release of the fact sheet from the 2023-24 survey, its use to derive fresh estimates of poverty has opened a new scene in the poverty debate.

  Poverty line

  To recap the key issues, we note that the conventional approach to measuring poverty is to identify the expenditure on a set of goods and services at which the basic human need is met in terms of an accepted calorie norm, calling this minimum expenditure the poverty line. The original line was drawn where the household expenditure on goods and services met the minimum calorie norm, and its subsequent revisions were based on updating it by using price inflators. In this methodology, the HCES plays a crucial role as it provides the distribution of households/persons according to expenditure classes. 

Studies have established that the different recall periods used in NSS surveys have a direct bearing on the estimated monthly expenditures that can, in turn, affect the position of poverty line and poverty estimates. This issue has now been laid to rest with the NSS adopting a modified mixed reference period (MMRP), where expenditure on food items consumed is recorded for a week, other regular goods and services consumed recorded for the last month, and a yearly reference period used for the remaining items.   

The Tendulkar committee, however, recommended moving away from anchoring the poverty line to calorie intake due to the absence of any definite evidence of a linkage between calorie input and health outcomes. Instead, it considered the urban basket, associated with the urban poverty level of 27.5, based on the uniform reference period (URP) of one month, used in the 2004-05 survey, as being generally acceptable, and proposed it for rural areas. It nonetheless computed the rural poverty line using the mixed reference period (MRP) data that record expenditures on a monthly and yearly basis. The committee, having used a new reference basket and newly derived price indices based on the same dataset, stated that its headcount ratio should not be compared with earlier estimates. 

The committee further recommended that as and when the MMRP was adopted by the National Sample Survey Office (NSSO), the MRP-equivalent per-capita expenditure should be shifted to MMRP, using the method adopted by the committee for shifting from URP to MRP per-capita expenditure for 2004-05. The Planning Commission estimated poverty rates based on the 2011-12 NSSO survey, but on the basis of MRP, not MMRP, and using price indices and methodology as suggested by the Tendulkar committee, though MMRP-based expenditure data were available. 

The methodology suggested by the Rangarajan committee was different from that of the Tendulkar committee. A food basket that met the normative calorie, fat and protein requirements defined the food component of the poverty line basket. The median fractile (45-50 per cent) values of clothing expenses, rent, conveyance and education expenses were treated as the normative requirements of the basic non-food expenses of clothing, housing, mobility and education of the poverty line basket. To these two were added the observed expenses of all other non-food expenses of the fractile classes that met the nutrition requirements. 

However, this report, unlike the Tendulkar committee report, was not officially accepted. Thus, we have the last official poverty estimate using the consumption expenditure data for 2011-12 using the MRP. This showed that 25.7 per cent of rural and 13.7 per cent of urban population were below the poverty line. As against this, the methodology used by the Rangarajan committee estimated 30.9 per cent of the rural population and 26.4 per cent of the urban population as being below the poverty line in 2011-12. 

Alternative datasets

  The withholding of the 2017-18 HCES survey for unspecified data quality reasons led to the use of alternative datasets and simulation of household consumption expenditure distribution based on questionable assumptions. Scholars used household consumption expenditure recorded in surveys like PLFS and from sources like Centre for Monitoring Indian Economy (CMIE). 

Experts like Surjit Bhalla generated synthetic expenditure distributions based on national accounts data. These exercises, however, took little notice of data comparability issues. Based on the 2022-23 HCES data and updating the 2011-12 Rangarajan poverty line using price inflators, Rangarajan and S Mahendra Dev also recorded  low figures of 12.3 per cent and 8.0 per cent in rural and urban areas, respectively, for 2022-23. The improved recording of expenditure with the introduction of multiple visits to collect data introduced from 2022-23 and the resulting non-comparability with earlier data however is not considered in these estimates. 

The SBI team has used household expenditure distribution based on the latest 2023-24 survey using the MMRP. The brief methodology noted in the SBI report says that the new poverty line has been adjusted for decadal inflation and imputation factor derived from the NSSO report using the 2011-12 (based on MRP consumption) poverty line estimate of Rs 816 in rural areas and Rs 1,000 in urban areas. Clearly, this method ignores the higher estimate of household expenditure resulting from the use of MMRP in 2022-23 and 2023-24, while using a markedly lower poverty line using the MRP data, available for the estimate of 2011-12. 

The other important point brought out in the SBI research paper is a decline in the gap between rural and urban poverty. This can be attributed to the identified urban areas in the survey in 2023-24 remaining the same as in 2011-12, except the possible inclusion of statutory towns, declared by the state governments. The urbanised parts of rural areas or the potential ‘census towns’ have not been included in the urban frame of HCES in the absence of any census since 2011. Had these potential ‘census towns’ been taken out of the rural frame, rural poverty is likely to be higher than that reported by the SBI research team. 

The new datasets have opened more dimensions in the estimation methodology. For one, this uses three monthly visits to households, instead of one, to collect the data. Second, expenditure data is available after inclusion of imputed values for the free provision of items like laptop/personal computers, tablet, mobile handset, bicycle, motorcycle/scooty, clothing (school uniform), footwear (school shoes, etc), besides free food items. Imputation for these items and their inclusion would further mark up the household consumption expenditure. 

Both these changes are expected to capture the expenditure much better but severely impact comparisons with the past surveys. This implies that the figures based on the new NSSO data for the years 2022-23 and 2023-24 would not be strictly comparable with those of 2011-12 as they underestimate poverty. This would be valid both for the estimates based on Tendulkar as well as Rangarajan methodology. The poverty debate after the release of the NSS 2022-23 and the current survey is thus unlikely to be conclusive and not different from the earlier great Indian poverty debate. 

P C Mohanan is a former member of the National Statistical Commission and Amitabh Kundu is Professor Emeritus at L J University, Ahmedabad