I am working on a background paper on the Democratic Republic of Congo (DRC). The DRC has conducted its last census in 1984, very few births and deaths are formally recorded, public services are in shambles - which makes that no-one can really say how many people live in the country and how they are spread. Most of the country's economy (many sources say 90%) is classified as "informal" - people growing what they need for their subsistance, selling this and that on the market, trading goods and services, digging for gold. A large part of the trade of the rich mineral resources goes unrecorded: for example, most of the country's gold is reportedly smuggled to Uganda, whose gold exports are far superior to the amounts Uganda can produce. No-one has a clear idea of the money that changes hands in illegal arms trading - only that it must be a lot, given the abundance of small weapons in the DRC.
Yet, when you visit the web-sites of eminently established institutions such as UNDP, OECD, the World Bank or the African Development Bank, you discover precise statistics on GNP per capita. The problem is, the data I find on those sites vary from US$ 85 to US$ 714. Just to make this clear: one organisation says, in 2007, the average gross national prouct of the DRC has been US$ 85 per citizen, while another source says, it has been US$ 714, and most situate it somewhere between 150 and 300 US$. If you're lucky, the web-site contains the raw data which these calculations are based on and you can check the underlying assumptions. It makes a huge difference whether you assume the current population of the DRC is 58 million or 68 million people; no-one really knows...
But does it make any sense at all to calculate GNP/capita based on official GNP statistics when an estimated 90% of CNP goes unrecorded? Then again, anyone who has been to the DRC will confirm that the GNP per capita must be abysmally low, at least for most citizens and in terms of purchasing power parity, as the manifestations of most dire poverty are ubiquitous. Or does the GINI coefficient of inequality need dramatic revision, to take into account of all those hidden riches concentrated in the hands of what must be very, very few people?
Precise statistics in utterly imprecise environments are to be taken with a huge ladle of salt.