# Elections and The Price of Gas

With the 2016 election season fully underway we will soon face the inevitable controversy of how elections affect gasoline prices. There are lots of *opinions* about how or why gas prices change before an upcoming election but no real objective analysis. Since gas prices regularly decline, on average, after the summer driving season, is there an election effect? That is, **do gas prices in the United States change by a larger amount leading up to an election than in other years?**

The answer is *no*. Elections have *no impact* on the price of gasoline.

Gasoline price data in the United States is recorded monthly by the U.S. Bureau of Labor Statistics. Their numbers go back to 1976, so it includes a sizable number of election years — nineteen, including midterm elections. It is important to adjust these numbers for inflation as this compensates for the price changes from the constantly decreasing buying power of the dollar — the Greenspan Effect, if you will. That is, a ten cent change in the price of gas in 1980 and in 2010 aren’t the same, and inflation-adjusted prices account for this. The adjustment is done using the Consumer Price Index data which is also available from the BLS. In this case, we’re using constant 2008 dollars.

We expect the price of gas to show some regular changes each year. Seasonal changes in prices and even employment numbers are common. Gasoline typically gets more expensive when the demand for it is higher in the summer driving season, and it falls off afterward. This is an important factor that needs to be a part of the analysis.

We can understand and account for this seasonality by breaking the data into parts. There are the parts that repeat annually (seasonal changes), the parts that change more slowly (trend), and the parts that change more quickly (remainder). Performing this additive decomposition gives us the results below. The slowly changing “trend” component is the largest. Seasonal variations swing about seventeen cents from high to low, but the remainder, the fast-changing fluctuations, can be in excess of a dollar, though they generally exist inside of a quarter of a dollar on either side of zero.

The seasonal component here is the part we’re interested in first. Not too surprisingly we see there is a 7.7 cent increase in the summer, peaking in the June driving season, and a 9.2 cent decrease in gas prices in the winter, with the bottom arriving in December. Interestingly, since elections are in November, they occur during the natural seasonal price decline.

**The key to determining the impact of elections on gas prices is to use the non-seasonal components of the prices for comparison.** That is, by considering only the trend and remainder components. Excluding the seasonal fluctuations removes the price decline effect that we see every year at the end of the summer driving season and lets us compare election and non-election years to look for election-related price changes.

I’m defining the time before the election that we are interested in as August through November. The last quarter before the election seems to be the time we should pay the most attention to. If there are any real trends in prices leading up to an election, they should be apparent in this time frame.

What we find when we look at price changes in this time frame is that only four out of the last 19 *election years* (1980, 2006, 2008, 2012) have shown a drop in gas prices (red bars on the plot below). And only two of those were by more than a dime. The average change here is a 2.3 cent drop, but that is heavily influenced by the 2008 drop of $1.40/gallon, statistically an outlier. The median value of an **8.1 cent increase** is more in line with the typical behavior.

And we find that just six of the nineteen *non-election years* (1983, 1989, 2001, 2005, 2011, 2013) showed August to November nationwide gas price decreases, and only one of those was more than a dime drop (blue bars on the plot below). The average price change in non-election years is a 3.6 cent increase with the median value of a **3.8 cent increase**.

This tells us that the non-seasonal median gas price change between August and November in an election year actually *increases* by 4.3 cents/gallon in an election year compared to the same time frame in a non-election year. This is perhaps easier to see if we look at the distributions of price changes.

Were it not for the two outliers in the election year data, the distributions of gas price changes would be nearly identical. So the question is: are the differences important. This 4.8 cent increase we found in the median value of a gallon of gas, is it significant? Does it *mean* anything?

To answer this we run a statistical test known as an *independent t-test*. The output of the test is the probability that differences between the two distributions are due to chance. What we find is that there is a 52.7% probability that differences between these two distributions are chance. Statistically, this is taken as the difference between these two sets is the same as zero.

That is, elections have no effect on gas prices.