If oil production shrinks fast, GDP impact models don't work: IMF.

Washington Post: Two IMF economists try to tackle peak oil in a new working paper, “Oil and the World Economy: Some Possible Futures.” (pdf) “Authors,Michael Kumhof and Dirk Muir don’t make any definitive predictions about how the oil supply will evolve. Two of their five scenarios:
– If, in fact, oil is much more important than many economic modelers have assumed, then the blow to growth from even a modest plateau in oil could be quite large—lowering growth rates by up to 1.2 percentage points over the next two decades.
– Oil production starts shrinking rapidly. This is the doomsday scenario. Some studies have suggested that global oil production is currently on a plateau and will soon start shrinking in by around 2 percent per year. Existing wells will dry up. The world will increasingly rely on oil from places that are more expensive to develop, such as Canada’s tar sands. What happens then? Nothing good. According to the IMF’s modeling, prices could increase by 800 percent over two decades. Growth rates in Europe and the United States would be reduced by at least a full percentage point—and much more if oil turns out to be more important than we thought. “Relative price changes of this magnitude would be unprecedented,” the authors note, “and would almost certainly have nonlinear effects on GDP that the model is not able to capture adequately”.”