A couple of things here leap out at me - the first is that we are relying on statistics for all this. I seem to recall there was a great saying about statistics . . .
Secondly, Graviman raised a good point with his comment about the aerodynamic efficiency of our cars. All the really great auerodynamic cars were built a while ago now - the DS19, the Mako Shark, Subaru SVX etc. Now we have less fuel-efficient bodies (less aerodynamic and also getting heavier).
One example is in the British Touring Car Championship - one team use the Honda Accord last year, and have moved to Honda's new Civic for this year which is less aerodynamic. So they had to get more power from the engine to make it as competitive.
The big problem I have with this thing about renewable energy, though, is that it isn't (renewable that is - on a macro or micro scale). There Ain't No Such Thing As A Free Lunch.
We put up wind farms to convert wind energy into electricity. This is taking energy out of our weather system. They also have a massive carbon footprint during manufacture and installation (new roads/cabling etc). The big ones in remote places can take up to 20 years to pay back their 'cost', and they are only supposed to last 25. AND why, when I pass the farms in the UK do I invariably see about a third of them not turning and generating? They'll take longer to pay off if they don't actually use the things.
Solar panels convert the sun's energy. This is energy that would normally go into the solid objects, warming them and enabling them to release this energy back into the air and warming it up after a cold night. This process also creates thermals which defines how our weather system changes.
Wave farms take enrgy out of our seas, potentially changing the natural currents. If you want to see massive localised climate change, try moving the Gulf stream away from Scotland - we might finally be able to get some decent skiing.
Once we add loads of wind farms and solar panels, all sorts of havoc could be unleashed in the weather systems of our plant.
Of course, this is only extrapolating from previously observed phenomena - but that's what everyone else is doing, right?