I understand that the IEA is trying to make a prediction for the future, so perfect accuracy cannot be expected. However, looking at the graph of World Oil Production 2002 - Present, there seems to be an increasing discrepancy between the various sources. Furthermore, the IEA's predictions seem be high-side estimates more frequently now than before.
There is a reason for this Light299. The IEA numbers are for 'total liquids' in just about the widest definition one could devise:
http://omrpublic.iea.org/currentissues/sup.pdf   (page 51)

Definition of Supply
In order to achieve a mass balance in the world oil supply and demand table (Table 1), supply includes not only crude oil and NGLs, but also various types of heavy oil-like hydrocarbons and natural gas-based, coal-based and renewable-based sources which are used as oil product equivalents and are included in our definition of demand. These non-conventional oils include other hydrocarbons and alcohols (including Brazilian alcohol fuel and those used in gasoline blending elsewhere), Canadian synthetic oil production, Venezuelan upgraded Orinoco extra-heavy oil and orimulsion, oil shales, South African coal-based and natural gas-based oil substitutes and methanebased blending components such as MTBE. Total supply of these products is estimated at around 1.4 mb/d in 2003. Refinery processing gains are also shown as a source of oil supply in Table 1. Care needs to be taken in reading the text and tables to distinguish between crude oil and total oil supply. Thus, in Tables 1 and 3, total oil supply is shown (excluding OPEC).

So, included are all such things as ethanol; sobering thought that, ne c'est pas? ;)

To elaborate: all non-conventional liquids are included, if these have been increasing in recent years (as, presumably, they have) one can draw certain conclusions about the production of fossil oil over that time.