Comment
Edward Glaeser: The connection between violence and cities is as old as the walls of Jericho. Historically, the threat of war caused people to cluster together behind city walls, because of the strength that came from numbers. The actual impact of war was often to disperse population, as people fled cities and returned to subsistence farming, when chaos challenged the secure property rights needed for urban commerce. In the wake of the September 11, 2001, attacks on the World Trade Center and the Pentagon, many observers suggested that a new threat of terror would act as a dispersing form, as people sought to avoid tall skyscrapers that were obvious terrorist targets.
Blomberg and Sheppard's fine paper represents an attempt to ask whether terrorism is indeed a force for urban dispersal or whether terrorism instead drives people together. Certainly, the theoretical case is ambiguous. High levels of density do seem to offer tempting targets to terrorists, and this effect should mean that an increase in terrorism disperses people. Two effects counter the tall targets effect: the costly travel effect and the city walls effect. The costly travel effect notes that terrorism also strikes at transportation modes, including buses and airplanes, and as the costs of transportation rise, people generally come closer together. The city walls effect emphasizes that there are scale economies in defense.1 London's system of antiterrorist cameras would have been much harder to implement in a lower-density environment.
Blomberg and Sheppard's paper addresses the theoretical ambiguity about cities and terrorism with two extremely rich data sets. They then connect this evidence with city-level evidence on urban expansion. They look across a large sample of cities and ask whether cities with more terrorism are more compact or more spread out. Since their specification has the logarithm of urban land use on the left-hand side and the logarithm of population on the [End Page 291] right, the regressions can be interpreted as asking whether terrorism increases or decreases the amount of land use per resident.
Before proceeding to their main empirical question about terrorism and land use, the authors provide an overview of the Memorial Institute for the Prevention of Terrorism (MIPT) data on terrorist attacks. Perhaps the most empirical fact that comes out of this overview is that terrorism is actually declining in most regions of the world, at least as measured by number of incidents. The decline of terrorism in Europe since the early 1980s is particularly striking. Latin American terror peaked somewhat later, but it has also declined. Terrorism incidents in North America have always been low, but they too have been generally falling since the 1970s.
The one region that has seen an explosion of terror over the past eight years is the Middle East. America's experiences with terrorism since 2000 have essentially reflected the ability of increasingly sophisticated Middle Eastern terrorists to export violence across the Atlantic. The authors have made good use of these data in other papers, and the data set clearly represents a major resource. The declines of terrorism in Europe and Latin America may provide clues about how to combat the new surge of terror in the Middle East. The time trend data also suggest that if we expect terror to have a major impact on urban form, we are more likely to see that impact in the Middle East than in Europe or the United States.
After showing the time series of terror, the authors give us some geographic pictures of terror within a few countries. In the United States and England, terrorist incidents are remarkably concentrated in the largest cities. English (or Irish) terror is strongly concentrated in London. American terror tends to be in New York, Washington, Miami, and Los Angeles. The connection between urban size and terror in these two particularly salient nations might lead us to think that the tall targets effect will dominate and terrorism will push people to decentralize.
However, Blomberg and...
