> Different regions have dramatically different average commute times which don't directly correlate with population. Thus disproving your point.

Average commute time is not the desired metric here, see Duranton and Turner[1].

[1]: http://www.brown.edu/Departments/Economics/Faculty/Matthew_T...