Khebab could look at it after he finishes the net export heavy lifting, although I suspect he will be taking a break after the slides are done.

Motivated by your suggestion, I started playing around a little with Excel, trying to create a simulation model. I didn't get very far! Maybe somebody here can get me on track.

I thought what I should do is to create basic supply and demand curves. How about a single supply curve, volume as a function of price. And then two demand curves, one for the internal demand of producers and then another for the demand of non-producers. Of course one could create lots of supply and demand curves, e.g. one for each major global participant.

What ties it all together, of course, is that the equations have to balance. There will be some global price at which the sum of all volumes supplied equals the sum of all volumes consumed.

So that basic equation would get solved for every time step by the simulator. One could then look at the resulting amounts consumed by non-producing nations or whatever other number is of interest.

The first challenge is to get nice shapes for the supply & demand curves.

I think WT's essential point regarding exports is that the elasticity of producers is significantly different than for non-producers. If the price goes up, more profits go into the pockets of producing countries. Surely with the higher prices being paid by non-producers, the producers will be happy to export more petroleum. But if the price goes up by 10%, maybe 7% of that just goes into the pockets of the producer, so the volume they consume goes down only as much as a non-producer's volume would if the price went up 3%.

My instinct is to make the supply-demand curves linear, so then the solutions are just linear too. I'm not sure if this is going to make things interesting enough or not.

The fun thing, of course, is how things change over time. The producing curve(s) should follow the Logistic equation or something similar. Should the elasticity just be some constant???

Then: what about time variation in the demand curves? The simplest thing is just to stick in some time-varying functions that are just assumed. But maybe there should be some real dynamics. E.g. if Qatar can spend an extra zillion dollars this year, maybe building a huge ski resort by the edge of the Persian Gulf, then that will drive up their demand next year - they'll want to be producing the snow for those slopes!

It's a nice little project! Unfortunately I don't know enough Java to be able to make a little applet out of this sort of thing. But if I get a spreadsheet put together, I'd be happy to post it.