An Attempt to Calculate The Price of Gas

[greylar]

With everyone wanting to know how much money they will save in gas with their new Prius I became curious as to what the price would be given a particular price per barrel of oil. What I found is that gas prices fluctuate pretty consistently with the price of oil and thus can be calculated with reasonable accuracy.
(SKIP THIS PART IF YOUR EYES GLAZE OVER AT THE MENTION OF MATH.)

The magic number it seems is around 42. That is the number that I came up by using the DOEs statistics. I took the price of gas between 2000 and 2007 $1.91 of which 48% was due to the price of crude at $39 per barrel ($1.91 x .48 = .9168), meaning that crude oil contributed 91 cents to the price of gasoline. Now take $39 and divide it by the .91 and you get 42.85. This is the factor that you can use to determine how much crude oil alone contributes to the price of oil and is NOT coincidently the number of gallons of oil in a barrel.

You might say refining oil is not a one to one ratio since you only get 19.5 gallons of gas from a barrel of oil. But they use almost every drop of oil to produce something and there is very little waste so the calculation would be more complicated but effectively the same (price of oil) x (percentage used for gas) / 19.5 or using the example above (($39 per barrel) x (19.5/42.85) )/ 19.5 = .91cents, the same number as being contributed to the price of gas above.

(END OF EYE GLAZING MATH SECTION)

The point being of all the math above is that the balance of the oil doesn’t go to waste so to make the math easier it is ok to assume that you get 42 gallons of gas from a barrel of oil and thus can divide the price per barrel of oil by 42 and get its contribution to the price of gas. Easy Huh? If you can't open the spreadsheet the basic numbers are $140/bbl = $4.51, $170/bbl = $5.23, $200 = $5.94

The attached spreadsheet assumes that taxes refining and distribution costs are fixed, (which we know they are not). But they fluctuate not so much with the price of oil but with inflation and the overall economy. So I think it is reasonable to assume that the final price of gas as calculated by my chart is more of a minimum number and it could be higher based on additional factors such as inflation or disruption in refining capacity etc.

One final note, oil is priced in futures contracts, which means that the oil price today will not be reflected in the price of gas for some period of time. There is some debate about this but I think it is safe to assume that it takes at least 2 weeks and probably more like 2 months for the price of gas to fully reflect an increase in the price of oil.

It is interesting that at the current price of oil, gas is under priced by .40 right now. If you don’t have MS Office download Open office for free.

My wife thanks you for letting me post this stuff here instead of making her PRETEND that she has any interest at all.

Thanks G

Reference Links:
EIA - Primer on Gasoline Prices
Oil to Gasoline

[qbee42]

Doing this cost calculation with one data point does nothing to show whether the calculation is valid at today's prices. To be useful you should include data points on either end and something in the middle.

Tom

[patsparks]

That is very interesting. Of course you do know that the answer to the question of life the universe and everything is, 42!!! See The Hitchhiker's Guide to the Galaxy. by Douglas Adams.

[greylar]

Originally Posted by qbee42

Doing this cost calculation with one data point does nothing to show whether the calculation is valid at today's prices. To be useful you should include data points on either end and something in the middle. Tom

Yes normally I agree but in this case the model works for almost any point in the last 10 years. In fact it works reasonably well for the last 20 years even without accounting for cheaper refining and distribution costs 20 years ago. So there is good reason to think it will work going forward. So if you have one data point that predicts something with relative accuracy then there is little need for more.

G

[greylar]

Originally Posted by patsparks

That is very interesting. Of course you do know that the answer to the question of life the universe and everything is, 42!!! See The Hitchhiker's Guide to the Galaxy. by Douglas Adams.

LOL I had forgotten that. We there you go no need for any corroborating evidence.
G

[Sheepdog]

So you are saying that 142 a bbl as it closed on Friday should mean that gas should cost what again? In Central FL right now today it is 3.89.

[greylar]

I think $4.56 as a national average, The national average is currently 4.08 which means if the price of oil stayed where it is the price of gas would go up about 48 cents. The attached spreadsheet lets you put any price you like and it will calculate my guess at the price of oil.

As I said before $200/bbl would be about $6

G

[morpheusx]

I am not sure if I follow all your calculations. I was told a few years ago from a co-worker who used to manage a few regional gas stations that they basically divide the price of a barrel of oil by 35 then adjust a few cents based off of competition in each market. I don't know where the 35 number comes from though. But that was also when gas was usually about 1.25 a gallon so I am not sure if it still holds true.

[greylar]

I am just diving the price per barrel by 42 and then adding in the price of the refining, and delivery. The math was to show that 42 is a valid number based based both on the fact that it works in the past and that I was able to get the same number when calculating backwards from the DOE numbers.

G

[qbee42]

Originally Posted by greylar

Yes normally I agree but in this case the model works for almost any point in the last 10 years. In fact it works reasonably well for the last 20 years even without accounting for cheaper refining and distribution costs 20 years ago. So there is good reason to think it will work going forward. So if you have one data point that predicts something with relative accuracy then there is little need for more.

G

If it works on many data points over the past ten years, then you aren't extrapolating from a single point. That's what I wanted to know.

Tom