Detroit

I have long been poaching Wendell Cox’s urbanized area data.  One good reason is that he carefully flags the parts that are core cities. 

Looking at these data for the decades 1950-2000, one can quickly correlate core city population growth with subsequent decade suburban growth.  Using just the 24 largest of these for which six decades of data are available, two types of lagged correlations are easily found.  What does one decade’s growth predict for the next decade’s growth, core areas vs suburbs?  Here are some correlations:

1950-60: core cities growth with subsequent decade suburban growth, r = 0.64; suburban growth with subsequent decade core city growth, r= 0.66;

1960-70: core cities growth with subsequent decade suburban growth, r = 0.18; suburban growth with subsequent decade core city growth, r= 0.39;

1970-80: core cities growth with subsequent decade suburban growth, r= 0.13; suburban growth with subsequent decade core city growth, r= 0.28;

1980-90: core cities growth with subsequent decade suburban growth, r= -0.01; suburban growth with subsequent decade core city growth, r = 0.17.

For each pairing, I italicize the larger value.  Strong suburban growth does a better job predicting strong core growth than the other way around.  These are mostly positive; growth in one part of the metropolitan area is not at the expense of the other. And core areas do well if their suburbs have been doing well.

Paul Krugman reports that “sprawl killed Detroit.” But strong suburbs usually have the opposite effect.

ADDED

Wendell Cox has more to say about this.