Monday, November 20, 2006

Sunil Jain on poverty in India

One of the better commentaries on the latest -- 'thick' -- survey from NSSO is from Sunil Jain. His column has several different strands, and I'm excerpting the poverty strand:

... In 1993-94, we're told India's poverty level was around 36 per cent, and this fell to around 26 per cent in 1999-00.

That is, it fell by around 1.7 percentage points per annum (it fell 0.9 per cent per annum between 1983 and 1993-94). In 2004-05, however, the NSS data suggest two poverty estimates, 22 or 28 per cent. Why there are two estimates is a long story related to how two types of questions were used for different samples, but suffice it to say the 22 per cent figure of 2004-05 poverty levels correlates with the 1999-00 levels of 26 per cent. That is, poverty levels fell by just 0.8 per cent per annum in a period when jobs growth rose anywhere between two and three times!

It is true, the country's poverty experts will tell you, poverty estimates are not calculated on the basis of wages, but from consumption data. But eventually, the two have to give the same results -- if wages go up, consumption will logically go up since the poor are too poor to save, and consumption cannot go up unless wages do, right?

Which corroborates the point economist Surjit Bhalla has been making all these years, that the NSS data capture less and less of the country's actual consumption. Consumption levels in the country can be got in two ways, from the National Sample Survey type of consumption surveys or from the National Accounts, which is where the GDP numbers come from, by aggregating value added in various sectors.

In the perfect world, the total national consumption should be the same from both methods. But, in India, the share of total consumption that you get from the NSS figures is consistently getting smaller -- in the latest round, the NSS consumption figure is less than half got from the national accounts.

I'm adding it to the reading list in this post.