Identifying BPL Households: A Comparison of Methods

The identification of poor households has been passionately debated in India. Since 1992 the Indian government has identified households as living below the poverty line (BPL) and hence eligible for certain benefits. Such identification exercises occurred three times, and a fourth BPL identification exercise is underway. Although the fourth BPL identification method aims to improve upon previous methods, the empirical implications of, and precise justification for, the revised method are not yet clear. This paper empirically examines the proposed Socio-Economic Caste Census (SECC) methodology and compares it with alternative proposals. Using variables in the third National Family Health Survey (NFHS), we show that the choice of a particular methodology (which may include exclusion criteria, inclusion criteria, and/or a scoring method) matters. These criteria - and even the exclusion criteria used alone - disagree as to which households are BPL; thus the criteria require empirical scrutiny and justification. We also visit the need for a scoring method to include sufficient indicators to match state poverty caps. Finally, we show how state-level BPL poverty caps vary if they reflect multiple deprivations in variables - such as malnutrition and housing - through a multidimensional poverty index, rather than reflecting expenditure-based poverty rates alone.