Thursday, November 08, 2012

World Bank's infuriating & inconsistent poverty data

The World Bank does some excellent development research, and really ought to be the first place you turn out for data on world poverty, inequality etc. As I've mentioned in an earlier post, however, its poverty and inequality data regarding India just do not add up. In that post I had to go through several hoops and construct complicated Lorenz curves to try and establish that things don't add up, but some latest figures will help establish my case much more simply.

Consider the numbers on this page : http://povertydata.worldbank.org/poverty/region/SAS

The % of Indian population at or below the various consumption thresholds are :

< $1.25/day - 32.7%
< $2.00/day - 68.7%
< $2.50/day - 81.1%
< $4.00/day - 93.7%
< $5.00/day - 96.3%

If we convert cumulative percentages to the group specific percentages, we get:


$1.25/day - 32.7%
$1.25 to $2.00/day - 36.0%
$2.00 to $2.50/day - 12.4%
$2.50 to $4.00/day - 12.6%
$4.00 to $5.00/day - 2.6%

Now assume that everyone in the respective groups earns/consumes the upper limit of the group.

(This is 'boundary condition' assumption that overstates the actual consumption of each group because in reality people in a given group will have consumption ranging between the lower limit and the upper limit of the group.)

Given this assumption, the total annual consumption of all the five groups combined comes out as follows


< $1.25/day - 32.7% ------- $179 bilion
< $2.00/day - 68.7%------- $316 billion
< $2.50/day - 81.1%-------$136 billion
< $4.00/day - 93.7%-------$221 billion
< $5.00/day - 96.3%-------$57 billion

The total is $908 billion. Keep in mind that this is the upper possible limit, the actual consumption by these five groups will be much lower. All of these figures are at purchasing power parity (PPP).

The data is from 2010, when India's private final consumption was Rs. 45 trillion. However, the world bank poverty lines are based on 2005 prices, so we divide this number by 1.4 (the price level went up by 40% in India in those 5 years) to get the relevant consumption figure - Rs. 32 trillion. (All figures from 2010 and for price level changes from the Reserve Bank http://dbie.rbi.org.in/DBIE/dbie.rbi?site=statistics)

At the then prevailing PPP rate of roughly $1 = Rs 20, this converts to total consumption figure of $1,600 billion. Which is to say, that the remaining 3.7% of India, the top 3.7% consumes atleast (1600-908)/1600 = 43% of India's total consumption! Not only is this figure unbelievable, the World Bank's own chart on consumption by quintile on the left of the same page (http://povertydata.worldbank.org/poverty/region/SAS) refutes it directly. The top 20% of India consumes 42% of our total consumption, so the top 3.7% can't possibly consume more than 43%.

Again, remember that the $908 billion figure is most likely a gross overestimate, so the comparison is even more stark than the one I show. The World Bank data page contains an arithmetic absurdity, one more egregious than the now irrelevant Ryan-Romney tax plan.

So what gives, World Bank? This is simple stuff, there should be consistency checks for this kind of thing on your pages, no? Unless the RBI data is seriously incorrect (which I doubt, and plus, both RBI and World Bank get their macro data from the same source - MOSPI in India), I think my calculations hold.

My sense is, the $1.25/day figure of 32.7% is quite rigorous and alright - it corresponds to and roughly matches with India's national poverty estimates - the percentages estimated to be living under each subsequent threshold are severely overstated. There simply is no way that 96.3% of India's population lives below $5/day (2005 prices), which in today's prices and after PPP would mean something like Rs. 150/day.

Hopefully, once the 2011 round of price comparison and consumption surveys of the World Bank is collated, we will have a much better view.

3 comments:

alternative investments retail investors said...

Having once worked at the WB, I can assure you that this is par for the course.

Anonymous said...

i thought this was well known to be a problem and not just an WB issue.

in particular, the poverty line figures come from household surveys (NSS) where as the private consumption figure you mention likely comes from the NIPA (national income accounts). and historically these have produced quite different pictures of consumption. i don't know if there is any progress on reconciling them

also, i think this is not particular to india but in many developing countries the national income accounts produce an estiamte of consumption that is much higher than that produced by household consumption surveys.

that household surveys miss some of the top income households is part of the explanation but i'm not sure what else is going on ...

Anonymous said...

i dont find it hard to believe that 80% of the country lives below 2.5$ a day; the issue might be the distribution amongst the top 20%, but I dont think that debate is as important as solving the situation where, using your upper-limit calculations, top 20% consumes >60% -hj

PS: your blog is always a great place to wander into every now and then, keep it up old boy!