All very good points, Andrew.
There's some other factor at work, however, which doesn't seem to be explained by either a classical supply-and-demand model or a commodities-speculation model. And that's what really interests me.
I managed to scrounge up a couple of charts tracking both the market price of oil and the retail price of gasoline over a six-year period. I re-scaled the data according to the big boom/bust cycle in 2008 which represented both the record high and record low of both products. This had the effect of making the rest of the chart track rather nicely between the two, with a couple of exceptions:
(crude oil is red, gasoline is blue)
See how there are three big excursions in which the price of gasoline experiences a bubble which is out of proportion to oil? I've taken two of those and blown them up:
In both of these cases, the price of gasoline rocketed upwards by a large amount just prior to a small increase in the cost of oil. Not only was this price increase predictive rather than reactive, but note especially the area which is highlighted in yellow. Both of these areas represent a steep decline in the price of gasoline (a correction towards pre-spike levels) which occurred while the market value of crude oil was still slowly increasing
If the retail price of gasoline were tied only to the market price of crude, then we would not see this phenomenon. And we've already established that consumer demand for gasoline is not sufficiently elastic to cause this.
This is where I get back to my "emotional" theory of Post #19. These anomalies represent periods of time in which the retail price of gasoline was controlled by hysterical speculation on the part of those involved in the process of creating and delivering gasoline, irrespective of the behavior of the commodities markets.