However, this is based on the national poverty line, the one produced by the Planning Commission and which varies from state to state and from rural India to urban India. The World Bank does not undertake surveys to collect data on household expenditure (not income). Thats still undertaken by NSS and the last large NSS large sample is for 2004-05. This shows a rural poverty ratio (percentage below poverty line) of 28.3%, an urban one of 25.7% and a composite one of 27.5%. Unless one argues growth doesnt trickle through to the poor at all, poverty ratios in 2008 should be lower.
However, there is also an international poverty line and let us restrict ourselves to expenditure or income poverty, as opposed to other measures of poverty. In 1985 prices, this poverty line was fixed at $1 a day per person. Obviously, if one uses different poverty lines, one will end with different poverty ratios, even if the same survey data are used.
In common citations of the $1 a day poverty line, people often forget it was conceived at 1985 prices and sometimes, a broader $2 a day poverty line was also used. De facto, the poverty line wasnt $1 a day.
It was $1 at 1985 prices, but $1.08 at 1993 prices and what Chen and Ravallion have now proposed is yet another revision to $1.25 at 2005 prices. If poverty lines are jacked up, naturally poverty ratios will increase. And indeed, poverty ratios dont increase linearly, because expenditure and income distributions tend to be log normal. Thats precisely the reason trickle-down benefits of growth lead to sharp drops in poverty ratios as the thick part of the distribution passes above the poverty line. Drops from 25% to 10% occur within a couple of decades.
But once the thick part of the distribution has passed above the poverty line, further drops to 5% or thereabouts become much more difficult. It is also because of this clustering around the poverty line that apparently dramatic increases in poverty occur if the poverty line is increased from $1.08 to $1.25 or even to $2.What complicates matters is that $1.08 or $1.25 isnt at official exchange rates, but at PPP (purchasing power parity) exchange rates. The difference between official and PPP exchange rates occurs because the prices of goods and services, particularly non-traded goods and services, are lower in developing countries.
Whats now happened with poverty figures is a natural corollary to an earlier revision of PPP output figures in China and India, especially the former. Since prices in China and India were higher than what was earlier assumed, Chinas PPP national income was reduced by around 40% and Indias PPP national income by around 30%. Following the same logic, the PPP poverty line is now proposed at $1.25 a day and the focus is again more on China than on India. While the logic is clear, there is still lack of clarity about robustness of the Banks PPP survey. For instance, using its own survey, ADB thinks the poverty line should be $1.35 a day. Using $1.08 a day, the World Bank estimated 879 million to be below the poverty line in 2004. If one uses $1.25 a day, the figure dramatically increases to 1.3996 million, almost 1.4 billion.
For India, the increase is from 266.5 million to 455.8 million. If one goes by the national poverty line, the poverty ratio is 27.5%. If one goes by the new World Bank measure, Indias poverty ratio is 42%. And if one goes by ADB, the Indian poverty ratio is 54.8%. Note that none of this negates four propositions. First, growth has reduced poverty in India. Second, global poverty has dropped because of poverty reductions in China and India. Third, again because of China and India, the world is on target for the poverty-reduction MDG (Millennium Development Goal) target. Fourth, the problem is in sub-Saharan Africa. However, two further questions remain to be asked. First, what do we do about poverty, regardless of how we measure it Second, lets assume the Indian population to be 1.1 billion. Then using 54.8%, 602.8 million Indians are poor and 497.2 million are non-poor. Note these figures are for individuals, not households. Do such figures gel with information on consumer expenditure They dont, even for non-elitist items like TV and bicycle ownership, LPG-usage and pre-paid mobile connections. Household expenditure surveys under-estimate consumption expenditure and fall short of consumer expenditure thrown up by national accounts. Unless that problem is resolved, numbers hide more than they reveal.
The author is a noted economist