Manipulation of prices in general have been a feature of post 2002 politics. I mark it at that time, because that was when the stock market became much less prone to price changes more than about 2.5%.

Based on data for daily market volatility, 1998-2002 were quite volatile as the tech bubble inflated and burst, but 1993-1997 were quite stable. At first glance, the data seems to say little more than that stock markets are volatile around economic downturns. That doesn't seem too surprising.

my principal objection would be that the price of gasoline is dropping when the actual supply of gasoline has dropped well below normal levels.

Gasoline supplies are 21 days, same as this time last year.

Moreover, oil prices have dropped about $40/bbl - $1/gal - since their July average; given that, how could gas prices not drop?

And yet, the way Saudi Arabia sets it price and quantity delivered to its customers seems more consistent with Saudi political decision making than simply maximizing dollar profits would imply.

Russia was much the same way during the Warsaw Pact days - profit was not the only consideration in where oil was piped, and what price was paid for it.

But that would imply that oil producers have a political agenda not actually coupled to monetary accounting, which would then lead to the idea that oil is not merely an economic good which the invisible hand distributes to its faithful servants.

Ask Chavez how much his various grandstanding oil shipments have cost, and he would likely answer that in a case like Cuba, the Cuban doctors have done more for his countrymen than the dollars supposedly lost by shipping oil at a 'loss.'

Mondays are always Monday unless you are looking across the International Date Line. So, when I take the weekly gasoline prices, which are are polled the same way whether there is a Federal holiday or not, they represent the first through 52nd (and on occasion 53rd) weekly values in any calendar year. Fridays are always Friday and so, with the exception of Thanksgiving Week and the various holiday weeks, we always have a 5-day average for that week from M-F. It's just as easy to pick a moving average of any suitable length between the WTI price (spot market) and the composite wholesale spot market price for gasoline.

Furthermore, the values shown are the rolling 4 week average value so the "error" of shifting days is extremely small. Besides the year-to-year preccessional effect on the movement of certain holidays that are fixed (e.g, the 4th of July) versus those that float (e.g, Memorial Day, Labor Day) are minimized AND the 4-week rolling average further reduces that effect. What is relevent is to pick a sufficiently representative period of time where short-term transients are minimized in the time-series, while teasing out the trends.

And the trend is most striking in 2004. I'm open to other suggestions as to what caused the 2004 decoupling, why it occurred right after the 4th of July and right before the Democratic Convention and then was sustained for the next 3 months before gasoline prices finally started rising (at a faster rate than oil).

Furthermore, the dollar effect is nowhere near what you suggest. If you take the four month period in 2004 (July through October) using the gasoline supplied from the EIA data you'd find that the total amount supplied was 1.13 billion barrels or 47.43 billion gallons. If the marginal cost differential was $0.10/gallon (which actually looks pretty close to the difference between the "normal wholesale price differential" and the price differential during this period of time), you'd only get a difference of 4.7 billion dollars. But with an oil aquisition cost of ~$74.5 billion and a wholesale value of the gasoline sold at around $60 billion (just by itself), not to mention the value of all the other products produced and sold, a differential cost of around $4 billion is a small price to pay to have voters happy and your favorite Congressional representatives back in power in addition to a friendly White House.

This cycle, we aren't happy in NC as we are still paying around $3.80-$3.90/gallon, when you can find gas. It does not seem to be much of a problem if you live along the some of the Interstate Highway corridors, but even in the state capital it is a problem. Yesterday and this morning, only 3 out of 12 gas stations along two routes I took had gas at some of the pumps. The rest have all their pumps shut off and the pump handles bagged. We are being told that this is likely to continue for another 1-2 weeks. Fortunately, I ride the bus.

I only have a couple of minutes to devote to this. Perhaps when my wife has recovered I can spend more time on this.

When I say that the gasoline price decoupled, I do not mean that it has remained uncorrelated. I mean the previous relationship can be shown to have been changed and at a very specific time.

However, if you had actually looked at the data and presented it differently, you would have discovered that the relationship between oil and gasoline prices was strongly correlated during the period of time from January 1 through June 30 with an r² of 0.965. After July 1 through Election Day 2004, that correlation went away as oil prices climbed and gasoline prices didn't (they declined until the very end of the period in October). In that period the slope of the curve practically disappeared and the r² fell to 0.265.

For comparison, the r² of the period from January 1, 2001 until June 30, 2004 was 0.798 and showed a high degree of correlation (and the path up versus path back down is also quite evident). The point is that something distinctly different happened in 2004.

After the election over the next 6 months, the correlation curve shifted (to the right) but returned to a similar slope as the period from Jan 1, 2001 until Jun 30, 2004 AND returned to a highly correlated state (r²= 0.878). Now if you think you can explain why this occurred, I'm all ears. Manipulation does not have to have occurred in every election year for it to have occurred once and be perceived that it occurs every time.....

Elesewhere, you challenge me on the 2007 issue, yet you devoted a portion of your blog on the dropping of the gasoline stocks and how sharp the decline was (would you like the reference?). This is a plausible explanation for the short-term (February-May 2007) for the correlation to "fail" once again. It's more difficult to make that argument for 2004 as stocks build slightly in July 2004 before falling. We also have refineries running flat out with very high utilizations (something else that was discussed in TOD) and those refineries did not get to that state by themselves.

I can provide a much more detailed analysis, but the main point is that the ability to manipulate gasoline prices to severely lower them is probably much more limited than in the two preious Presidential elections.

As a final note: you should have noticed that the price in your quote $2.17/gallon was not possible as, until two days ago, this was less than the wholesale price of gasoline on the spot market. Last time I checked, they are still paying federal and state gasoline taxes in Illinois.

There was one brief period of time where gasoline prices rose rapidly in 2007 to look as though the two would once again recouple. That period from Febraury to the end of May 2007, reversed and went back to the decoupled (but mirroring), post-2004 operating curves. If memory serves, gasoline stocks were low and declining during this period as there were a number of discussions on the minimum levels here on TOD.

Note that the same type of "price manipulation" trend is present in the 2002 data, it just is not as pronounced as in 2004. I could similarly argue that the trend is more pronounced in the two presidential elections since 1998 than it has been in the midterms. But I also look at the data going back to 1990 (where my weekly data ends and my monthly data begins because I have not been willing to go to the Federal Repository library and dig this data out of the files) and similarly conclude that there does not appear (from the way I'm handling the data against a baseline of 1992) to be manipulation prior to 2000. But, I can say that in 1973, that there was an big perceived difference between gasoline at $0.269/gallon and 0.399/gallon. Perhaps a relative value that is contemporaneous to the year in question.

Also note that this is not what I was setting out to discover when I started putting together this project more than 5 years ago AND in 2004, it was fascinating to watch the numbers evolve the way they did and a completely new pattern to emerge in near real time.

Here is what the 2007 data looks like when you remove the other years and just delalwith the calendar year (note the scale change)

One argument could be that it looks, sorta, kinda, like 2004. But the sharp rise and then what looks on the other graph as a sharp fall in gasoline prices is not so pronounced when you zoom into the data. And more importantly, we know what what caused the rapid rise in gasoline prices relative to oil prices.

http://i-r-squared.blogspot.com/2007/04/this-week-in-petroleum-4-25-07.html

If it wasn't for this, I suggest that the price rise of gasoline would have been more in line with the oil price rise and not the curves we see here.

Still hunting for a reason to explain 2004, so when you find one in the production, stocks, or other petroleum related data, do let me know. SAS hasn't found it yet.