Some readers may have noticed that I have repeatedly visited these themes: (1) cities always have and always will decentralize; (2) doing something about it is neither desirable nor feasible; and (3) the whole discussion requires analysis at levels of spatial detail for which we do not have a lot of data.
In this light, have a look at Brookings' recent "Job Sprawl Revisited: The Changing Geography of Metropolitan Employment" This link links to the PDF for the full study. (HT to Ken Orski). The work updates earlier work by Glaeser, Kahn and Chu which pioneered the use of zip code data on job location. Even zip codes can be too large to capture what is going on, but let's not quibble.
The new study reports (gasp!) lots of "job sprawl" in America. Years of industrial policy ("smart growth") have only had measurable impacts on the cost side -- witness the effects on reduced housing affordability.
But what is odd about the report is that the author seems not to like her findings. Before the findings are introduced, the author does the obligatory tour of all of the (alleged) problems that "job sprawl" brings. Skip that and go straight to the findings.