OK, Manhattan in hand … here’s an analysis that improves on the first one. I’ve taken the ratio of [gasoline price x gasoline quantity] to median household income. The gasoline quantity data are from Table 5.13a of the DOE’s Annual Energy Review 2004, in thousands of barrels.
Table 5.13a reports “estimated petroleum consumption, residential and commercial”. That means that I am overstating the amount of residential petroleum consumption, which at least will bias the results in the direction I want; if I’m overstating residential consumption, I’m overstating residential expenditure, so if it declines, then we know that in truth the decline is even larger than depicted. The analysis would be more precise if I could break them out. But here it is anyway:
If anything, this decline is larger than just the ratio of gas price to median household income. Interesting.