It was kindly pointed out to me last week by erich, a commenter on my article at Energy Bulletin that the data set I had used had not been adjusted for inflation. I had originally argued that it shouldn’t make much difference. I thought that as I was dealing with the price changes within a year and as the ANOVA treats each year as a separate group the effect of inflation would be negligible. However after thinking about it for a while I decided it would be worthwhile to run the inflation adjusted numbers to remove any doubt on the conclusions I had drawn from the original data set. Using this consumer price index (CPI) information I adjusted my original data set to 2010 dollars and reran the ANOVA.
Figure 1: Statistically significant Bonferroni-Holm test results looking at the difference between years in the monthly change in price for crude oil (US dollars per barrel), simple average of three spot prices; Dated Brent, West Texas Intermediate, and the Dubai Fateh, August 1982-June 2012. Data from http://www.indexmundi.com/commodities/?commodity=crude-oil
Figure 1 shows there were less results rejecting the null hypothesis, 32 compared to 59 in the unadjusted data which goes to prove erich’s point that at least some of the effect I was seeing was due to inflation. I feel somewhat vindicated at the same time that my original conclusions still stand. As erich said in his comments after I shared the new results “Adjusting prices for inflation makes your analysis and conclusions more robust and defensible” and so I thank erich for first raising the issue in a polite and encouraging manner.
I also tracked down some more oil price data sets that I hadn’t come across before and I was excited to see if my hypothesis, that oil volatility has significantly increased since conventional oil production plateaued in 2005, could be replicated.
This morning I came across an article on the 2012 United States Presidential “Energy Election” at a blog called Con Carlitos, written by Calvin Sloan. He had an interesting footnote which explained something I had not quite grasped and I will quote here in full:
As noted by Steve Austin at OilPrice.com: “Brent is the real international benchmark.Two-thirds of the oil consumed in the US is Brent, and two-thirds of international crude is priced to it. (Brent crude is sourced from fifteen oil fields in the North Sea.) Still the media and market persist in quoting WTI, rather. This is a US singularity, like the non-adoption of the metric system.”
Due to this I decided to look specifically at Brent crude prices as the strongest conclusions can be made from that data. I looked at both monthly and weekly Brent Crude prices between May 1987 and June 2012. Using average annual CPI data from the U.S. Bureau of Labor Statistics I adjusted the monthly and weekly Brent crude prices to 2012 dollars .
The monthly data returned only four results rejecting the null hypothesis. 2008 was significantly different to 1993, 1994 and 1995 and 2011 was significantly different to 1995. I was surprised this result was so low and intrigued to see if the weekly data was any different.
Figure 2: Statistically significant Bonferroni-Holm test results for weekly Brent crude prices May 1987 to June 2012.
It would appear from Figure 2 that the monthly effect was largely masking the difference in Brent price volatility between years. When looked at on a weekly level we see that 76% of the statistically significant results occur after the 2005 production plateau.
It can be concluded that while oil prices may be highly volatile on a week to week basis the effect is reduced on a monthly basis. We can also conclude that post 2005 we have seen a marked increase in weekly oil price volatility. The first six months of 2012 data is the fourth (equal with 2010) most volatile year for oil prices in the last twenty-five years behind only 2011, 2009 and 2008. The data from July and August 2012 was not included in this data set but it is my guess they would push 2012 further up the ranks. It is even possible by the end of the year that 2012 could be the second most volatile year after 2008. I for one will be watching oil prices over the next few months with great interest to see if the great yo-yoing continues.