I had previously cited Bumoo Lee's study (here) of commuting in 2000. He looked at where commuters worked (downtowns, sub-centers or dispersed) for the 14 largest MSAs in 2000. Of these 42 average trip times, the shortest commutes were for workers driving to dispersed job sites in Phoenix. Call it sprawl.
There are good commuting data from NHTS 2009. These are the most detailed, offering summaries of individual records for MSAs for four metropolitan subdivisions: "urban," "suburban," "second city," "town and country," as well as the "metropolitan" area as a whole.
Consider NHTS' three major trip types: journey-to-work, home-based shopping, and home-based recreational. Consider also average trip time for each area and purpose as well as the corresponding variances. There are, then, as many thirty readings for each major metropolitan area (three trip types, five area designations, two statistics each). There are fewer than thirty readings for some MSAs where observations are too few (I ruled out cases of N less than 100).
I looked at the 14 major MSAs that Lee studied. Which of these had the most of 30 possible readings better than than US 50-metro area average? You guessed it. For Phoenix, 29 of the 30 readings beat the 50-area overall. No other metro area came close. Call it sprawl.
This is all about the co-location of people and their destinations. Co-location was most successful in most places for shopping trips. Where was co-location least noticed? In Washington, DC. As far as worktrips go, Washington's major employer is the least footloose. The traffic consequences are clear.
Head-to-head comparisons between Phoenix and Portland (not one of Lee's biggest 14) were possible for only four summary statistics because of small NHTS sample sizes for Portland. For journey-to-work metropolitan area average commute times, Portland outperformed Phoenix, 22 minutes vs 24.6 minutes; for the corresponding variances, Phoenix wins, 263.1 vs. 414.9. For metropolitan area shopping trips, Phoenix wins, 11.9 minutes vs. 14.4 minutes. For the corresponding variances, Phoenix wins again: 91.5 vs. 154.7. But in 2009, Phoenix had a population almost 70 percent greater than Portland. Good for sprawl.
I must say that the cost comparisons ignore the benefits side and are therefore incomplete. But the "high costs of sprawl" have been the popular drumbeat for years.