Inferring Economic Activity from Gasoline Use?

If so, consider the gasoline intensity of GDP:

Figure 1: Log gasoline supplied to real GDP ratio, in (thousand bbls/day)/(4xbn Ch.2012$/yr) (blue), NBER defined peak-to-trough recession dates shaded gray. Gasoline supplied adjusted by Census X13, log transform, X11 seasonal adjustment. Source: EIA, BEA, NBER, and author’s calculations. 

Gasoline supplied is decreasing by 5 ppts faster than real GDP growth per annum (in log terms). Or, the ratio of gasoline supplied to real GDP is decreasing by 5 ppts per year. Moreover, given structural changes in American shopping and commuting habits over the past two years, I would be particularly wary of relying on changes in gasoline use or vehicle miles traveled as a way of inferring economic activity.


24 thoughts on “Inferring Economic Activity from Gasoline Use?

  1. Moses Herzog

    It is one of many tools useful for casually gauging inflation. For GDP……. I can’t say “useless” but…… I sure as hell can’t see it, because there are too many things gasoline is used for just for subsistence. And honestly I don’t know that people even cut back that much on their recreational driving even when prices rise. I would say if the typical American lowers gasoline consumption (i.e. driving) when prices are high they only adjust their driving behaviors moderately But that’s all anecdotal on my part, so……..

  2. Macroduck

    An interesting irregularity shows up across the three recessions represented here. In the first two, gallons per unit of output rose. In the third, gallons per unit of output fell, massively at one point. So in non-pandemic conditions, we go about our non-GDP business, using gasoline to go to the park or visit Aunt Martha, as well as to work for the most of us who keep our jobs, while output falls. In quarantine conditions, we all stop driving to visit Aunt Martha and we stop driving to work. Lots less gasoline how into GDP and into non-GDP.

    I see hope in these data. We reform our behavior when our lives are threatened. Well, commence reforming, people.

  3. pgl

    “given structural changes in American shopping and commuting habits over the past two years, I would be particularly wary of relying on changes in gasoline use or vehicle miles traveled as a way of inferring economic activity.”

    Everyone knows changes in relative prices have effects on people’s choice. Everyone except Princeton Steve of couire.

  4. David S

    We need to see the time series on this graph going back to 1820—at the very least 1880.

    On a personal note, I delayed filling up my car by a few days this week because I’ve been noticing that my local gas station has been dropping prices by about 5 cents a day. I like to think that I contributed to a decline in monetary velocity—if that’s how that works. Speaking of velocity, the idiots who drive these monster pickup trucks haven’t been slowing down on the highway at all.

    1. Menzie Chinn Post author

      David S: Did gasoline exist in commercial quantities in the 1820s?

      EIA doesn’t report gasoline product supplied prior to 1991. If you have a source, I’ll be glad to try it…

    2. Willie

      Funny thing about that, though. Even those absurd bro-dozers get better fuel economy than a 1966 pickup with a straight six and two speed automatic. I would be surprised if gasoline use wasn’t declining as a percentage of GDP. There are more cars, sure, but they are far more efficient than the equivalent 20 or 30 or longer ago. That doesn’t take into account the slow gains in electric and hybrid cars. Over the last five years, the number of electric cars I see on a regular basis has increased dramatically. Maybe that’s normal across the nation, maybe it’s just my neighborhood, but we aren’t that unique.

  5. rsm

    Why fetishize economic activity as if it is the goal? Shouldn’t economic activity be seen as incidental to more important things like pursuing happiness even if it is noneconomic and irrational (falling in love)?

      1. Moses Herzog

        I was thinking about BlueStatesKopits’ death count numbers for brown people the other day and wondering how much money I could make with an app and a web address titled “” ?? Future goldmine??

        I notice BlueStatesKopits has zero thoughts about Kentucky flood death count numbers and the amount of federal aid offered. Strange……..

    1. Macroduck

      For the many students of formal rhetoric who frequent this comments section, we have a gem here!

      “Why fetishize economic activity…” is a perfect example of “begging the question” in its original meaning. The point rsm means to argue is that we fetishize economic activity. However, rsm doesn’t want argue that question directly because it’s not demonstrably true. So rather than debating the question, rsm “begs” it by inserting the assumption of fetishization into a different question.

      “Begging the question” is cheating, trickery used by those who don’t have a leg to stand on. They stand on a soapbox, instead.

      1. AndrewG

        There’s that. There’s also the fact that economic activity is generally good in and of itself. The level of “economic activity” can imply higher quality transactions. More options lead to better matches/choices. (I think this is where price indices are really important over time, so we can compare real values, say, across countries.)

        This is putting aside situations with bad incentives of course. We put way too much carbon in the atmosphere due to our economic activity, for example.

  6. Erik Poole

    Good points.

    Do we have any kind of estimate of the net change in gasoline and diesel consumption brought about by increased use of online shopping?

    Has the online shopping phenomena reduced fuel consumption as well as climate emissions and if so by how much?

    1. AndrewG

      Great questions. Thing is, online shopping still requires delivery. So we’d have to take that into account somehow. But presumably Amazon is more efficient at delivering things to you than you are (energy-wise). Then again, deliveries require more packaging, which would add weight/bulk.

      Related to this is how big box stores and grocery stores have fared – two very car-intensive operations. Maybe gas stations can tell us something.

    2. Macroduck

      Excellent question, for which answers have not yet been definitively reached. Pre-Covid studies were few, and tentatively reached the conclusion that delivery generated less CO2 than brick-and-mortar shopping. The conclusion was tentative because authors recognized the complexity of the behaviors being studied. For instance, there is evidence that the number of trips to brick-and-mortar shops decreased less than the number of goods purchased – people could have purchased some items without additional CO2 generation simply by picking up more stuff, but ordered stuff for delivery, generating more CO2. So on a per-item basis, CO2 production is less for online items (maybe), but accounting for all behavior, the combination of online and brick-and-mortar shopping shopping may increase CO2 production.

      That’s before accounting for the surge in rapid delivery. We may assume that rapid delivery results in increased CO2 production, but I haven’t found any hard evidence.

      Finally, the swings in travel – to work and for recreation – since the arrival of Covid messes up any top-down effort to assess the impact of fuel use on various types of shopping.

      Good question. No good answers. Very messy.

      1. AndrewG

        ^ Macroduck’s is the non-lazy version of my comment ^

        “For instance, there is evidence that the number of trips to brick-and-mortar shops decreased less than the number of goods purchased – people could have purchased some items without additional CO2 generation simply by picking up more stuff”

        This is cool. Evidence that all the effort stores put into thinking about where to physically place products is probably worth it. Maybe changes to the layout/design of websites/apps just have lower returns (fewer spur-of-the-moment or serendipitous purchases to squeeze out).

    3. Baffling

      There is a large push to make final leg of delivery with electric vehicles. That will impact the use of gasoline as a proxy. This will be a big deal within two years. Already impactful.

  7. Erik Poole

    Thank you AndrewG and Macroduck.

    I looked for studies for about 25 minutes before I posted. I kept looking afterwards and found two studies:

    Shopping trip or home delivery: which has the smaller carbon footprint?

    Environmental Analysis of US Online Shopping

    I also saw a reference to a Carnegie Mellon Green Design Institute study but was unable to find the study referenced. Apparently the study concluded that online shopping resulted in 35% less fuel consumption and emissions than the traditional retail model.

    I agree Macroduck. Very messy. If for no other reason than that shopping behaviour either online or in a traditional bricks and mortar retail outlet can vary considerably. Presumably if one bicycles to the traditional retail outlet, the fuel consumption and climate emissions are minimal.

    Online retailers have an annoying habitat of often insisting upon shipping non-dangerous items to the residential address when shipping to the postal box address would use much less fuel.

  8. Ivan

    Temporary inflation spike caused by supply chain disruptions have eased as those disruptions, predictably, have been reduced. Well, wouldn’t you know.

    1. AndrewG

      “Well, wouldn’t you know.”

      There’s absolutely no doubt in my mind that if the Fed board in 2021 (who were saying what you’re saying) could redo 2021, they would. They would start raising rates in the fall.

Comments are closed.