Who are these 250 million people? The question shouldn’t be ducked. Nor should we avoid the question by suggesting all SC/ST/ OBC/Muslims are poor and all non-SC/ST/OBC/Muslims are non-poor. There is a double problem with any surveys on poverty, the NSS included. First, data become dated. Second, and more important, these are surveys, not censuses. Knowing there are 250 million poor people isn’t of much use, unless we can ascribe names, faces and addresses to these numbers. This identification has also been a problem when some states have undertaken their own surveys on poverty. So we resorted to decentralised identification through surveys during the Ninth Plan (1997-2002), especially for rural India. A family was poor if its annual family income was less than Rs 20,000, it didn’t own a TV or fridge and its land-holding was less than 2 hectares. The Tenth Plan (2002-07) refined this to 13 parameters for rural India (land-holding, type of house, clothing, food security, sanitation, consumer durables, literacy status, labour force, means of livelihood, status of children, type of indebtedness, reasons for migration) and seven parameters for urban India (roof, floor, water, sanitation, education level, type of employment and status of children), with states like Kerala, Haryana and Maharashtra adding their own variations.

Such decentralised identification based on ownership of physical assets (and similar variables) sounds eminently sensible, particularly because panchayats are involved. But it hasn’t worked that way. Consider how the system works. On each parameter, you get a score. Each parameter has a weight and the scores are thus aggregated to obtain an overall BPL (below the poverty line) index value. No matter how decentralised the survey, this is bound to be perceived as a top-down approach, since choice of parameters, choice of weights and formulae for aggregation are all inherently subjective decisions. For every method suggested, an equally plausible alternative is possible. This is the stuff one writes academic papers with, not use it for framing public policy on such a sensitive issue, where a double mistake is possible — identifying poor as non-poor and identifying non-poor as poor. It is always best if one can figure out some method of self-identification. Old people belong to a different category. Of the 77 million old (more than 60) people in India, 25 million have been estimated to be BPL. With overlap across the two categories, there are 34 million widows.

... contd.