If there is anyone who understands international comparison of incomes, it is Angus Deaton, who has computed comparable price ratios for every nation going back to the 1800s. In this year’s AEA address, he considers the recent revision of global poverty by the World Bank, which added 500 million to the rolls of the ultrapoor. The address is principally a technical discussion of how PPP ratios are created, and the problems this creates when attempting to examine poverty and inequality rather than (as originally intended) GDP ratios.
At present, prices are collected in various countries, incomes are converted using the derived PPP, at those PPP the average level of national poverty lines in a number of countries is taken in international dollars, and then that international poverty line is compared to national income distributions, with each step causing problems.
It seems a more direct method would to define a standard of absolute poverty in terms of goods – say, X amount of a nation’s stable cereal, housing of X square meters in the most common local style, X clothing per person (cotton T-shirts strike me as common from Burundi to Belgium), X amount of some common local produce, etc. The cost of that bundle in local currency can then be compared to the local currency income distribution to reveal the absolute number of impoverished in each country, and this can simply be summed to calculate global poverty. The extra data collection work seems fairly limited, the definition of poverty in goods – what we care about in the end – is by definition exact in each country, and there is no need to even attempt to compute general cross-national PPP ratios, whose use for comparing, say, GDP, is clear to me, but whose use for computing international poverty is less clear.
http://www.princeton.edu/~deaton/downloads/presidential%20address%2019january%202010%20all.pdf (Draft version – Final version in AER 2010.1)