Skip to content

Time trends in gasoline prices

Dan Lakeland posts this graph:


I’d take the y-axis down to zero and reduce the number of tick marks (0, 5%, 10% on y-axis and every 10 yrs on x-axis). Also, why start at 1980? Why not go back to 1940? Some data are here:


I never think about gas prices because I only rarely drive a car. But I’ve heard that it’s considered a big political issue right now. My thoughts on this issue are stuck in the 1970s. I was shocked to see an ad recently of some hybrid car that had highway mileage in the 20s. Back in the 70s there were cars that got over 40 miles to the gallon! I can understand that there are gas guzzlers even today, but I was surprised that a carmaker would actually be advertise such mediocre mileage.


  1. derek says:

    I'd keep the number of tick marks the same (every 5 yrs), but halve the number of tick mark *labels* (every 10 yrs).

  2. Nick Cox says:

    I don't understand the widespread enthusiasm for writing y axis numbers vertically, a common feature of giraffe graphics. Here the numbers 0.02, 0.03, … could easily be horizontal, making reading more comfortable.

    (Even better: just use a percent scale and write 2, 3, etc.)

  3. Daniel Lakeland says:

    Thanks for the comments. Being an engineer I often want to read fairly exact values off of graphs, so I prefer my graphs to have denser tick marks, so the R defaults were fine with me.

    I was also mainly interested in differences so zero didn't seem that important, but I can see your point there. It's interesting to know for example that current prices are about 2x what they were in the dot com era, and seeing zero would show that automatically.

    The main reason I only went back to 1980 is that the most convenient source of gas price data only had data that far back. I looked all over and like most government organizations, they're pretty awful about publishing their data in a conveniently analyzed form. They have weekly data that goes back to not very far, and then yearly data that ends in 2006 and blablabla.

    If I were planning to write a paper on this I'd get more data on gas prices, and try to get more data on average fuel economy, and pollution output, and average commute length (census has this I think).

    Still, with dead simple analysis, you find out that automobile fuel, and associated taxes, is around 5% of per capita GDP, and that tells me a lot more than OMG $4.55 WTF!!! which is essentially what you hear in the media.

  4. John S. says:

    I had a car in the 80s that got 45 miles to the gallon (a Geo Metro). As I understand it, the reason cars no longer achieve such efficiency is because of safety regulations: today's cars have to be a lot heavier to meet crash standards.

  5. Andrew says:

    I wish they were all so light they would bounce off pedestrians in a crash–that would be the safety regulation I'd prefer!

  6. Auto makers don't "advertise" their fuel efficiency; they are required by law to measure it (by a standard test under city and highway driving) and to list it on the sticker.

  7. Ben Bolker says:

    Current versions of the
    data that were used in the LBL report
    above are available at:

    I don't know how they dealt with leaded vs
    unleaded etc. — it would be interesting to
    see how much work it would take to recreate
    their graph exactly.

  8. Phil says:

    John, no safety regulations aren't to blame for the bad gas mileage, or at least that's not the main reason. The fact that there still are some small, light cars out there (like some Hondas and Toyotas) shows that they're still allowed to make 'em that way.

    No, the problems are that (1) people keep choosing larger cars and (2) people keep demanding higher performance. The average 0-60 time has gone from over 14 seconds (in 1981) to under 10 seconds. It takes a big engine to make that possible (not many people buy cars with 4-cylinder engines anymore), so that's one issue; also, since people have the capability to accelerate faster, they do so.

    If cars were designed with 1980 weights and accelerations, the fleet average (for cars, excluding "light trucks" which means most SUVs) would be 40 mpg now, not the actual 28 or so. Safe cars could be made at that weight by using more expensive materials (as has been done for bicycles) if people were willing to pay for it. A quick search finds a relevant presentation at

  9. Paul says:

    Anyone else find the title of the graph ambiguous? At first read, not knowing the context, it appeared to me that the graph was trying to tell us about the perils of time-travel.