The Glaeser/Redlick paper I discussed yesterday really is worth a read. It makes two very interesting points about the benefits of density:
The returns to skill are also higher in denser areas. As density doubles, skilled workers earn 5,400 dollars more relative to the less well-educated. This effect is quite statistically significant. One interpretation of this fact is that density and skills are complements; this is also supported by the propensity of high human capital industries to locate at the center of urban areas.
And this:
[W]e show that there is also a significant interaction between area density and individual skills. As density doubles, the college premium appears to rise by about 4.6 percent.
Neither result necessarily means that density itself increases the returns to skills and education. Density is correlated with high property values, and property values are high when a location offers high returns to labor. But the correlation between density and the returns to skill may also reflect the increasing returns from positive spillovers; workers tend to be more productive when they are close to one another.
In either case, restricting density negatively impacts welfare. The restrictions either they shut some people out of high-return locations, or they dampen the increasing returns to scale offered by a denser environment.
I do wonder, though, whether these statistical associations would hold if run using weighted density rather than standard density. Hmm.