Gasoline Affordability Reconsidered

Last Monday, the Los Angeles Times published an op-ed written by Indur Goklany and me about gasoline prices.  Yesterday, it ran in the Minneapolis Star Tribune Today, that same piece has been posted at the Christian Science Monitor and it will appear in their print edition tomorrow.  Our argument: Once you adjust gasoline prices in 1960 for both inflation and changes in per capita disposable income, you find that gasoline prices today are actually more affordable than they were back then.  Faithful Cato@Liberty readers might well recognize this argument given that it was first offered in a blog post here a few days back by Indur Goklany.

While the predictable grousing on the newspaper comment boards followed (hell hath no fury like a motorist who thinks he was told to stop whining about pump prices), some commenters raised a legitimate issue: Would the picture change if we used median per capita income rather than mean per capita income in our analysis?  Well, yes.  But not by that much.  Let’s walk through the numbers.

First some background.  Income data come from two very different sources.  Disposable income data are produced by the Bureau of Economic Analysis (BEA), an arm of the U.S. Department of Commerce, as part of its effort to estimate the gross domestic product (GDP).  Data on family and household income come from surveys conducted by the Census Bureau.

Disposable income per capita or mean disposable income is simply total disposable income divided by the population of the United States.  Median disposable income data, however, are not available because the GDP data do not come from household surveys.  Only surveys allow us to rank order all the households (or families) and find the number that divides the bottom 50 percent from the top 50 — the definition of the median.

Median income estimates from Census data (the Current Population Survey or CPS) are available only for households and families.  Data regarding median household income are only available from 1967 to the present, so the only measure available to us for longer term analysis is median family income.  But BEA and CPS definitions of income differ.  In 2001 for example, BEA personal income totaled $8.678 trillion while CPS money income totaled $6.446 trillion.   The two income time series differ in important ways.  For example, BEA data include property income and adjustments for underreporting of proprietor’s income.  

With that out of the way, let’s get to the numbers.

(Leaded) Gasoline prices in 1960 averaged 31.1 cents per gallon.  Median family income in 1960 was $5,620.  In 2006 (the most recent year for which we have reliable data), median family income stood at $58,407.  If the price per gallon were the same percent of median family income in 2006 as in 1960, the 1960 price would translate into $3.23 in 2006.  Unfortunately, the (median family income) data aren’t yet available for calculations applying to 2007 or 2008.

A complete history of fuel prices for 1949-2006, adjusted for changes in median family income, can be seen in the figure below.

Offsetting the fact that the price of gasoline as a function of median family income is probably (somewhat) higher today than it was in 1960 is the important fact that vehicle fuel economy is better today than it was then.  The gasoline consumed by passenger cars in 1960 was 14.26 miles per gallon.   By 2006, it was 22.4 mpg.   Even the mileage for all other 2-axle 4-tire vehicles (lights trucks, etc.) in 2006 was 18.0 mpg — higher than the fuel efficiency of cars in 1960.  Hence, the cost of the fuel necessary to drive a mile might well be less today that it was in 1960 if we’re adjusting for changes in median family income.  Again, I say “might” because 2007 and 2008 data are not yet available to provide concrete numbers.

On the other hand, it is certainly true that people have responded to higher incomes and better fuel efficiency by driving more.  Vehicle miles per capita in 1960 were 3,249; in 2006, it had increased to 9,171 (the numbers are author calculations based on vehicle miles traveled data from National Transportation Statistics table 1-32 and population data from Statistical Abstract of the United States Table 1. The 2006 VMT figures includes passenger cars and other 2-axle 4 wheeled vehicles).   Vehicle miles traveled is a function of individual decisions about where to live, where to shop, and how to spend discretionary income.  Many people, of course, made decisions about those things when fuel prices were at their historic lows (the late 1990s) and now find that those decisions are now more costly.  Adjustments are and will continue to occur on this front.

A comprehensive measure of how these various factors — higher fuel prices, higher incomes, better mileage, more miles traveled — work to affect the cost of driving is the percent of disposable personal income spent on gasoline and on all user-owned transportation expenditures over time.  And what do you know?  The percent of income we spend on transportation has been remarkably constant over time even though the distance we travel per capita has nearly tripled (The data for this calculation come from the GDP data available from the National Income and Product Accounts Table 2.5.5 line 69 (total user-owned transportation expenditures) and line 75 [gasoline and oil expenditures] and table 2.1 line 26 [disposable personal income]).

  • In 1960, gasoline expenditures were 3.3% of disposable personal income.  In 2007, gasoline expenditures constituted 3.4% of disposable personal income.
  • In 1960, total user-owned transportation expenditures were 10.8% of personal income.  In 2007, those costs constituted 10.5% of personal income.

So no matter how you slice the (available) data, it tells more or less the same story.  All things considered, the cost of driving is reasonably affordable today relative to what it has been in the past.

Update: Data links added.