How Have Different State Populations Changed Their Gasoline Consumption?

May 21, 2019

The behavior of motorists in the U.S. has changed a lot in the last 35 or so years, and these changes have led to a noticeable drop in per-capita gasoline usage in the 21st Century. But this decline has not been even across all states, and if Congress ever gets the Highway Trust Fund self-sufficient again and based largely on gasoline taxes, these discrepancies in state gasoline usage patterns could eventually have an effect on highway funding debates.

Background: Last week, I was having a discussion about how Texas is now the only highway “donor state” remaining in terms of the dollars of tax payments attributed to Texas by the Federal Highway Administration and deposited in the Highway Account of the Highway Trust Fund exceeds the dollar amount of total apportionments and allocations the state receives from the Highway Account. (The fact that the latter amount includes payments supported by the $110 billion in general fund moneys transferred into the Trust Fund over the last two decades, while the state tax payment numbers do not include general fund payments, explains why there aren’t more donor states in dollar terms.)

In explaining what was odd about Texas, I mentioned that they are credited with an outsize share of the diesel fuel and trucking excise tax payments into the Highway Account because they have a huge population, a big economy, a lot of miles of Interstate and other NHS roads, and all that NAFTA truck traffic that then makes its way to the rest of the Lower 48.

But gasoline consumption is still the primary determinant of Highway Account collections, and gas consumption is the result of billions of small decisions made by consumers every week. I wondered if there was something about the behavior in of consumers in some states that differed markedly from the behavior of consumers in other states.

The annual FHWA Highway Statistics series publishes Table MF-226 which attributes “Highway Use of Gasoline, by State” all the way back to 1949. I correlated those numbers with the Census Bureau’s annual July 1 population estimates (since 2010 from Table NST-EST2018-01 at and then going back to 1981 from Table 12 of the final, 2012 edition of the late, much-lamented Statistical Abstract of the United States, which is as far back as I could find Excel files). The Table MF-226 gallon numbers are then multiplied by the HTF Highway Account gas tax rate to get the estimated Highway Account gas tax payment numbers used for the donor-donee calculations and other tabulations kept by FHWA.

First, the nationwide totals. Total gallons of gasoline for highway use is the gray line, from Table MF-226 (orange line, right axis), and then the MF-226 totals for each year are divided by the Census Bureau population estimates to get per capita totals (gray line, left axis).

In terms of the big picture, 1981 was the tail end of the Iran oil shock (in terms of fewer barrels of oil being delivered to the U.S., the Iranian revolution had a bigger effect than the 1973-1974 OPEC oil embargo), and there was a divot around the time of the Gulf War in 1990. But the overall trend lines went in the same direction until per capita consumption peaked in 2004 at 466 gallons per U.S. resident. After that, the post-Katrina oil shock caused the initial changes in consumer behavior (dumping big SUVs in favor of smaller vehicles), after which came the first CAFE fuel economy increases in a generation in George W. Bush’s second term and a significant expansion of those CAFE increases under Barack Obama.

But those are national totals. I then wanted to pick a few individual states for comparison – Texas (obviously) but also a few others, large and small, with different characteristics. In terms of diversion from the current national average, here are the 12 highest and lowest per capita gas users (the 2017 national average was 416 gallons per person):

States With the Highest and Lowest Per Capita Highway Use of Gasoline in 2017
Highest Usage Lowest Usage
State Gal./Yr. State Gal./Yr.
Mississippi 549 Dist. of Col. 143
Wyoming 544 New York 275
South Carolina 537 Hawaii 310
North Dakota 533 Alaska 334
Alabama 523 Rhode Island 335
South Dakota 513 Washington 359
Missouri 497 Illinois 361
New Hampshire 493 Oregon 369
Tennessee 488 Utah 369
Delaware 488 California 371
Texas 487 Pennsylvania 373
Kentucky 477 Nevada 381

For reasons to be explained at the end of this article, I chose not to use D.C. or high-area, low-population states like Wyoming or the Dakotas in the long-term trend charts.

First, just Texas and the other vast mega-state, California.

Texas started out this period consuming 22 percent more gasoline, per capita, than the U.S. average. By 1991 this was down to around 10 percent over the national average and has stayed fairly consistent since, until the Lone Star State began to diverge from the mean in 2012. They are now up to 17 percent above the national average of per capita gasoline usage, almost where they started out in 1981.

California, meanwhile, began this period almost exactly at the national average but by the mid-1990s had been able to moderate its gasoline usage to stay 5 to 7 percent below the national average. This trend in per capita reduction has accelerated in the last decade to where California is now 11 percent below the national per capita average.

These trends have a lot of factors – population density, the availability of mass transit, unemployment rates, commute distances, the kinds of vehicles people purchase, etc.

Next, let’s look at the six largest states by population. In particular, New York stands out because the mass transit usage in New York City is so large (over 40 percent of all transit ridership in the entire U.S. is in NYC). (We had to lengthen the vertical axis to get NY to fit.)

Large population growth concentrated in urban areas can lead to divergence. In 1981, four states (Arizona, Colorado, Mississippi, and South Carolina) were fairly close to each other in per capita gas consumption (457, 481, 456 and 468 gallons per capita, respectively), and all were a bit above the national average. And in 1981 the AZ + CO population was 5.8 million, almost identical to the AL + MS population of 5.7 million. What a difference sixteen years makes. In 2017, the AZ + CO population was 12.7 million compared to 8.0 million for the AL + MS combination, and while the two Southern states have climbed to the #1 and #3 spots on the gas consumption per capita list, the two Rocky Mountain states have had their per capita gas use parallel the national average or go slightly below it.

The two states that have trended farther in the opposite direction than any other states sit together in the alphabetical list – Nevada and New Hampshire. The growth of Las Vegas has made Nevada, which has the ninth-lowest population density of any state, resemble an urbanized area in many ways. Meanwhile, we are not at all sure what the transformative factors in New Hampshire have been (changes in demographics of the population, greater integration with the Massachusetts economy?).

For better or for worse, the estimated amount of gasoline used on highways in a state is still the chief metric used to determine the amount of highway funding most states receive each year (explicitly for Texas for the past two years and implicitly baked into the formula shares for many states under SAFETEA-LU that have been largely frozen since 2009). If the Trust Fund is ever made solvent again based on highway user taxes, rates of gasoline usage will once again loom large over funding distribution debates, so more study of how these numbers are attributed – and whether or not they create “perverse incentives” for states – should be conducted.

Why some states were not included in the charts. 

High-area, low-population states have noisy data. The smaller the state population is, the more random noise fluctuations there appear to be in the per capita data. I started out by wanting to use Wyoming as a test case because its per capita gasoline consumption has always been among the highest in the U.S. But Wyoming, Montana and the Dakotas have a problem in that the Census numbers show large volatility in some of the year-to-year population estimates. As it turns out, in a state with a little over a half-million permanent residents, the influx and outflux of tens of thousands of oil and mining workers in boom times and slowdowns can have an outsized impact on the denominator of the per capita equation.

Likewise, Alaska and Hawaii are both unique and don’t really compare well with the Lower 48. Alaska is just too big – too big to have roads, in many instances, so a lot of intrastate travel is via commuter planes and general aviation. Likewise, in Hawaii, a much larger share of intrastate travel has to take place via air or water, because, you know, islands.

The D.C. dataset may be broken. I also wanted to use the District of Columbia as a test case because it is the only completely urbanized city that nonetheless is treated as a state for data reporting purposes. But the Federal Highway Administration’s dataset may be broken in the case of D.C. FHWA entitles Table MF-226 “Highway Use of Gasoline, by State” and that implies that it is based on the number of gallons used on the road in the state. But the data is collected from state fuel tax administrators and is based on the number of gallons sold and taxed in the state (with some adjustments), not the number of gallons actually used in the state. And in terms of the number of gallons sold and taxed in the state, D.C. has a unique problem.

It has been well-documented that, as D.C.’s high rate of economic redevelopment continues, there are fewer and fewer gas stations within the boundaries of the District. More and more commuters are being forced to gas up in Virginia or Maryland before or after making their daily trips into D.C., and more and more D.C. residents who live close to the borders of Montgomery or Prince George’s County, Maryland are having to drive across the line to buy gas.

According to Table MF-226, the most recent peak in the number of gallons credited to highway use in D.C. was 2000, with 169.2 million gallons. By 2017, that number had dropped to 99.5 million – 41 percent fewer gallons. At the same time, the Census estimate of the D.C. population in 2000 was 572 thousand, and that population has risen to 696 thousand in the latest estimates, an increase of 22 percent. Divide one by the other and per capita highway use of gasoline as determined by FHWA has dropped in DC from 296 gallons per year in 2000 to 143 gallons per year in 2017 – a drop of more than 52 percent.

Does this sound plausible? Let’s cross-check with VMT. Sadly, FHWA doesn’t appear to publish annual state-by-state VMT data in the Highway Statistics series but the annual tables (Table VM-3) show 3.5 billion miles of VMT in D.C. in 2000 and 3.7 billion miles of VMT in 2017, an increase of 218 million miles. Divide the total gallons from Table MF-226 by those VMT numbers and you get the average miles per gallon in D.C. rising from 20.7 mpg in 2000 to 37.3 mpg in 2017 – at the same time when the U.S. average mpg by that same calculation (Table MF-226 divided by Table VM-3) shows the total average fuel economy in the U.S. rising from 21.5 mpg in 2000 to 23.9 mpg over that same time. Not really plausible, especially when transit ridership in D.C. is declining. (Also, anyone who ever takes Uber or Lyft in DC has to note how few of those Uber and Lyft cars have D.C. plates.)

It seems more likely that reliance on state sales tax data for modeling highway use of gasoline in a state or state-equivalent may not work for the District of Columbia anymore. (Hat trip to Eno’s Paul Lewis for first noting the incongruity in the numbers.)

Search Eno Transportation Weekly

Latest Issues

Happening on the Hill