Nice that you and Frederik agree!
But please note eventually that I had said similar things as you (please see above). That is I wrote that I don't know what ILO is using, but that I think that they probably use the first of the discrete approximations. I used Taylor approximation (for the case of the GDP, please see above) to display how the last two formulas differ.
Please note that there is also a discussion about this page [here](http://johncarlosbaez.wordpress.com/2011/03/18/energy-the-environment-and-what-mathematicians-can-do-part-1/#comment-4955).