Back in March, I wrote about economists Matthew Kahn and Ed Glaeser's work comparing carbon emissions among metropolitan areas.
Among other things, they estimated how carbon emissions from driving vary across metropolitan areas. The variation in gasoline use is just as interesting, though. Kahn and Glaeser did not explicitly calculate this, but I did the simple arithmetic to convert their estimates of carbon emissions to estimates of gasoline use.
In my last post, I was working from their relatively short, "popular" article, which provided data for just ten MSAs. I have now put my hands on their full paper[1] (gated, unfortunately), which provides estimates for 66 MSAs. I've again converted their estimates of carbon emissions to estimates of gasoline use.
What's neat about Kahn and Glaeser's approach is that they've tried to make an apples-to-apples comparison. Simply calculating gasoline use per household is misleading because some metropolitan areas are wealthier than others, and wealthier people tend to use more gasoline, all else being equal. Also, households tend to use more gasoline as they increase in size; thus, a metropolitan area with lots of large households will use more gasoline than a metropolitan area with lots of one-person households. Kahn and Glaeser have tried to fix this problem by estimating how household income and size affect gasoline consumption. They then estimated gasoline consumption for a household with 2.62 members earning $62,500/year.
They also estimated the impact on gasoline consumption of MSA density, census tract density, and distance from the Central Business District. And they found that gasoline use varies across regions; Southerners simply drive more than people out West, for example, even after controlling for things like density. Using these estimates, they predicted, for each MSA, the "standard" household's gasoline consumption.
The results are below. Just for fun, I compared gasoline consumption in Austin to gasoline consumption in other MSAs. Austin does not fare well under their methodology, mainly because it is not very dense and a large percentage of the population lives relatively far from the CBD.
(Here's the spreadsheet on Google Docs)
Note that this allows a rough estimate of the relative amount of pain inflicted by gas price increases. For example, using their estimates, a $1 increase in the price of gasoline costs the standard household in Houston $60 more per year than the standard household in Portland. (Remember that these are aggregate MSA estimates; they don't permit comparisons between the residents of Houston's suburbs and the residents of central Portland. They also have not explicitly factored in the effect of transit on gasoline consumption, although that will be captured to some extent by their adjustment for density.)
[1] Kahn & Glaeser, The Greenness of Cities: Carbon Dioxide Emissions and Urban Development August 2008 NBER Working Paper #14238 www.nber.org.
