A little while ago I wrote about the relationship between crude oil prices and the price Sydney motorists are paying for petrol at the pump. The Australian Automobile Association (AAA) has now released their price data for June and, not surprisingly, prices continued to track moves implied by rising crude oil prices. The simple regression model suggested that average prices would be up 8 cents/litre. The AAA data shows a rise of 10 cents/litre in the average Sydney price.
Sydney Petrol Price Model (Brent Crude)
What is also evident in this chart is the rapid fall in crude oil prices over the last week. One explanation for this fall was an unexpected rise in US inventories. Other commentators have pointed to a self-correction mechanism with higher prices leading to changes in consumer behaviour which are anticipated to be reducing demand. These developments should mean that Sydney drivers will see prices stabilising after several months of steep price rises. That is, of course, unless the price falls are short-lived.
My original post was syndicated on the Oil Drum, where it has attracted a number of comments. One commenter, Anawhata, mentioned comparisons with Tapis (Asia) crude oil prices as opposed to Brent (Europe/Africa) prices. I chose Brent because it is the most widely quoted in our media, but using Tapis prices in the regression does give a very slight improvement to the model (R2 is 97% rather than 96%), For completeness, here is the chart for the Tapis model, although it is hard to see much of a difference.
Sydney Petrol Price Model (Tapis Crude)
Of course, an even better model can be obtained by using refined rather than crude oil prices. The chart below shows the result of modelling Sydney petrol prices in terms of Singapore 97 Octane gasoline (the R2 is 98%)
Sydney Petrol Price Model (Singapore 97)
UPDATE: Maybe I spoke too soon about crude oil prices! The lastest news is that a tropical storm heading towards Mexico and more difficulties with Iran on the nuclear question have started pushing prices up again.
Posts Feed
July 22, 2008 at 3:51 pm
Prawny –
On the radio this morning someone (it was too early to remember the name) mentioned that the recent crude oil price drop would take 7-14 days to filter through to the pumps due to storage tanks, current supplies etc etc. In contrast, my impression is that the lag between crude oil price increases and Sydney petrol price increases would be better measured with a stop watch than a calendar.
Is there enough sensitivity in the data to measure the time differences (if any) of your correlation with respect to the price trend (upwards or downwards) ?
July 22, 2008 at 4:04 pm
Unfortunately the data I have for petrol prices consists of monthly averages (from AAA), which is too coarse to measure time lags over 7-14 days. The crude oil (and Singapore refined data) is daily, so if only I could find daily retail petrol price data…
July 22, 2008 at 4:07 pm
Has anyone averaged the Perth prices from their FuelWatch scheme ?
July 22, 2008 at 9:51 pm
Daily historical petrol price data is sold by, for example, Informed Sources:
http://www.informedsources.com/capabilities/pricewatch.htm
Of course, if you are patient and persistent enough, you can harvest it yourself each day from, for example:
http://www.shell.com.au/petrolpricing_scuba/suburb.asp?state=NSW
July 23, 2008 at 10:05 am
Interesting to see that the NRMA is calling for petrol prices to be down to $1.35 based on falls in Singapore prices. Economists are saying the fall will not be as large as that. As you can see from the Singapore chart above, I’m with the economists: Singapore price falls would seem to point to prices around $1.50 not $1.35.
July 23, 2008 at 10:34 am
@Mark L: thanks to the pointers to the data. Now to work out how to automate harvesting that Shell data!
@Duncan: In 2007, The ACCC published a report on Australian Petrol Prices which included an analysis of Perth’s FuelWatch:
“This included an econometric analysis indicating some reduction in relative pricing margins between Perth and the eastern capitals in the time following FuelWatch’s introduction”
The econometric analysis was released in May 2008 and concluded:
“From the econometric analysis, on a conservative basis, the ACCC can say that there is no evidence that the introduction of Fuelwatch in Western Australia led to any increase in prices and it appears to have resulted in a small price decrease overall.”
This report also includes the fall in weekly average prices of 1.9 cents/litre which has been bandied about widely (e.g. here). However, the report itself notes that FuelWatch may not be the only factor at play. For example, they note “the entry of Coles into Perth was an event that may have had a price impact”
Overall, reading the ACCC documents these are fairly meek conclusions, and furthermore, a draft paper by a La Trobe econometrician discussed over on the Possum Pollytics blog suggests that there are weaknesses in the ACCC methodology. The paper argues that even their weak conclusions are hard to support using the price data.