.jpg)
I always like to take on conventional wisdom. Anytime someone says something like "you know that gas prices will be higher in the summer, they always are..." I just can't help but question it and take a look to see if it is true. So this is one I have started to hear lately and I have never really looked at the data myself.
Here is a plot of U.S. regular unleaded gasoline prices since 1991 (data courtesy of the U.S. Department of Energy). The plot has a log scale of course.
The gridlines mark January of each year. What do you think? Are gas prices higher in the summer? Hard to tell from this plot. I can pick out many confirming and non-confirming years.
Analyzing Gasoline Price Seasonality
As you have seen me do with other time series analyses, here it is useful to look at the monthly difference or % change each month rather than look at the original data. This is absolutely necessary for data that has such a strong trend or cycle over time (called non-stationarity). If we look at the % change each month we can then look at what the average % change is for all the January's, February's, etc. Here is a plot showing that:
Interestingly there appears to be some seasonality here. You will see that March, April, and May appear to have gas prices increasing, the summer they are flat, and Oct, Nov, and December show prices decreasing.
And for the fellow stat-geeks, it does appear to be statistically significant from a simplistic ANOVA look (p-value <0.000).
Typical Gasoline Price Seasonality
The above % change per month can be hard to visualize and still leave you wondering what does it really mean for me the gas buyer in my 4-wheel drive Hummer? So here is that result put into a typical year of gas prices for a year where January starts out at $2.00 per gallon:
So as you can see above, for this hypothetical year where gasoline prices start out at $2.00 per gallon in January, you would expect them to be in the $2.25 to $2.30 range in the summer months (about a 14% increase over January). Coincidentally, this year in January prices were at $1.77 so this simple seasonality would suggest prices would be $2.02 in the summer. Prices are already above that. But keep in mind, all the above is the average seasonality over all the years since 1991, any given year can be different, let's look a bit more into that.
How Much Could Gasoline Prices Go Up and How Likely Is It?
If we look at all the years since 1991, there were 15 out of 18 years where the price in June was higher than the price in January. So there is an 83% chance that gas prices in any given year will go up June vs. January. As far as how much, the average as shown above is a summer increase of 14%, but there is a 20% chance that prices will increase more than 24%. But only a 5% chance that they will increase more than 34%.
So I would summarize by saying that in a typical year gasoline prices will increase about 14% June vs. January. They might stay even if you are lucky but they have a fair chance of going up 25% but probably not more than 35%.
I haven't tried to fit a forecasting model to this as of yet, this analysis is just a historical/seasonal analysis to see how prices typically act. In the future, I will try to fit a statistical forecasting model more precisely and see what it says for the near future. I wonder about trading possibilities with an ETF like UGA.
Finally, for me personally, this analysis totally surprised me. My natural skeptic figured this often quoted common wisdom was a fallacy. Turned out I was wrong.
Update, here is another post looking at trading this seasonality and UGA.
...
4 comments: