Suburbanization has, for a long time, been a trend based on consumer preferences and larger trends, notably rising wealth and transportation and communications improvements (including the highways Baum-Snow investigates). Jackson (1985) finds U.S. suburbanization began at the end of the 19century. Indeed, he refers to “streetcar suburbs.” The 20th century U.S. experience is shown in Figure 1, which shows the percentage of US population living in metropolitan areas, and breaks that percentage down into central cities population and suburbs. The growth of the suburbs relative to the central city is seen well before 1950. Moreover, in the figure the relative decline of the central cities is understated because central cities have been annexing suburbs for many years.
The simple broad narrative is that, by and large, suburban living expanded throughout the twentieth century. Around the world, as incomes rise, people choose the mobility of the automobile; they overwhelmingly prefer the range and choice of personal transportation. As they choose automobility, origins and destinations disperse; and as these disperse, the attraction of the auto grows. It is a self-reinforcing cycle that is facilitated by better highways. But as with most public sector infrastructure developments, these usually follow rather than lead.
This article offers some striking facts. Before the interstate highway system, the percentage of the U.S. population living in suburbs went from 7.1 percent in 1910 to 23.3 percent in 1950. From 1950 to 2000 it had a smaller proportional increase, namely from 23.3 percent to 50 percent.
The central city of Copenhagen reached its population peak in 1950 and by 1990 had lost nearly 40 percent of its population; that is comparable to some of the highest losses in the U.S. Rust Belt. The central city of Paris reached its peak population in 1920 and has lost one-quarter of its population by 1990.
There are many other interesting points in this piece. I am not suggesting that highways do not matter, but the extent of the influence is maybe not as large as many people think.