We really need to step back and ask ourselves how we know what we know and why we believe what we know.
Consider Tanzania's gold production. This is a small example of a large challenge.
Our friends at Wikipedia give The British Geological Survey's estimate of 2006 Tanzanian gold production at 39 tonnes. Our American friends at the U.S. Geological Survey say the 2006 figure is 47 tonnes.
Here are their Tanzania figures 2006-2010: 2006, 47; 2007, 40.193; 2008, 36.434; 2009, 39.112; 2010, 42. (The rounder numbers tell me "estimators at work" and "no definitive data.")
Now let's look at Reuters; consulting government sources for 2011, it says "Tanzania's gold production rose to 40.4 tonnes in 2011 from 35.6 tonnes a year ago [2010] ..."
Reuters is talking off Tanzanian official sources and is saying 2010 saw 35.6 tonnes. The U.S. Geological Survey says 42,000. We're seeing U.S. estimates 20% higher than British and Tanzanian estimates.
Not that Reuters can even paraphrase a government press release correctly. Notice the source government statement says (emphasis added):
" ...the export volume increased to 40.4 tonnes from 35.6 tonnes recorded in 2010," the central Bank of Tanzania said in a report on its website.The folks at Reuters have confused the export figure with the production figure. Trust them only as far as you can throw them.
After all this numerology, you deserve a payoff. This is a pretty good one. From Mining Weekly, 26 July (emphasis added):
According to [Tanzania Minerals Audit Agency CEO Paul] Masanja, small- and medium-scale gold miners produce about 20 t a year, yet less than 2 t is recorded by government.We have these terrible give-or-take 20% production/export figures or estimates. The gold that falls between these statistical chairs is a very big deal, especially if we extrapolate the error rate to worldwide production. Think about the implications if worldwide production figures are off 20%.
It gets better. Masanja is saying (if I read him right) that the Tanzanian official export figures, cited by Reuters (in the example above) embed 2 tonnes of small/medium output in the total of 40.4 tonnes while not counting an additional (estimated) 18 tonnes.
Tanzania exports 40.4 tonnes of gold. Another 18 tonnes is knocking around somewhere. Where? Dubai? In jewelry? In central bank vaults?
Do you see the implications for our estimates, models, frameworks? We have gross variations in "official" numbers compounded by even bigger variations factoring in undocumented production.
