I did a quick, fairly unscientific analysis of the retail prices for a gallon of gas versus the underlying crude oil prices. Here is a graph, what do you see?
Just looking at the data, it would appear that retail prices are most closely aligned to the cost of refining rather than the cost of crude. The R^2, which measures how much of the variability of one set of data can be accounted for by another set of data, shows that crude prices and refining costs independently account for about 57% of the variability, so they both matter.
If you look closely at the data, and run more careful analysis of the numbers, you will see that pump prices increase immediately with spikes in crude oil prices and refining costs, but decline much more slowly as crude 0il prices recede.
In the first price hike in this data, which begins Feb 7, 2005 and extends to April 4, 2005, crude oil and refining costs rise an average of $0.054 per week and pump prices increase a similar $0.049 per week.
The decline, which runs April 4th to June 9th, 2005 (with crude and refining costs bottoming out sooner than June 9th), shows crude and refining declining at a combined $0.067 per week, while pump prices decline at $0.022 per week, or about 60% more slowly! It’s similar in other price run-ups and declines.
In those differences in slope is profit. Historic profit. Oh, and also, oligopoly pricing.
Sean Foote's Blog 