The riots that began in London have spread across the country. Decades of social-science research has delivered insights about these outbursts of violence, but hasn’t explained why they erupt when and where they do.
We do, however, understand how they usually end: with overwhelmingly force that clears the streets.
Riots are more common in democracies, and the U.S. has had its fair share. The deadliest was the 1863 Draft Riot. More than 120 people were killed when the streets of Manhattan were taken over by protesters, many of them immigrants, who were furious at the prospect of having to fight in the Civil War.
In the early decades of the 20th century, cities such as Atlanta and Chicago were torn apart as whites attacked newly urban blacks for perceived transgressions. Chicago’s 1919 riot began when a child crossed an invisible racial barrier while swimming in Lake Michigan. In the 1960s, there was widespread unrest. In many cases, including the 1965 Watts Riot, the violence began with an argument over law enforcement.
These public disturbances are a classic example of tipping-point phenomena, which occur when there is some positive feedback mechanism that makes an activity more attractive, or less costly, as more people do it.
There is a tipping point in rioting because the cost of participating -- the risk of going to jail -- gets lower as the number of people involved increases. If I decided to start rioting tomorrow in Harvard Square to express my outrage at the closing of the beloved Curious George children’s bookstore, it’s a pretty good bet that I would be immediately arrested. But if thousands of others were involved, I’d probably get off scot free. The police would be overwhelmed, and my probability of incarceration would fall to zero.
Thus, riots occur when the shear mass of rioters overwhelms law enforcement. But how do these mass events get started?
In some cases, such as the New York Draft Riot, organizers get people out on the street. In others, such as the 1965 Watts Riot, a peaceful crowd provides cover for initial lawlessness. Sporting events, such as Game 7 of the Stanley Cup Finals in Vancouver this year, can easily produce the crowds that allow a riot to start. Most strangely, riots can follow an event that creates a combination of anger and the shared perception that others will be rioting. The acquittal of police officers in the Rodney King case seems to have created these conditions in Los Angeles in 1992.
The London riots appear to have had a simpler starting point. About 300 people gathered at a police station to protest the shooting of a 29-year-old suspect. Once there were so many angry people in one place, setting fire to an empty police car became a low-risk piece of pyrotechnics for the protesters.
After riots, there is often an attempt to explain the outburst as the result of large societal forces. The events in the U.K. have been blamed on growing inequality and the current government’s austerity program. The disorder in the U.S. in the 1960s was attributed to racism.
But across U.S. cities, there has never been much of a link between unrest and either inequality or poverty. In fact, the riots of the 1960s were actually slightly more common in cities that had more government spending. Riots were significantly less common in the South, where the Jim Crow laws were making their long overdue exit. This isn’t to say that many people involved in riots don’t have valid grievances, but plenty of people have serious grievances and don’t riot.
Somewhat paradoxically, even though the police often provide the flash point for these outbreaks, larger police expenditures per capita in a city in 1960 was associated with fewer arrests and arsons when riots occurred. Even if a riot provides a wakeup call for police reform, in the short run, the outbreaks typically end only when there is enough law enforcement to ensure that such behavior leads to arrests.
I hope the U.K. can handle its violence with a purely police response, but in the U.S. restoring law has typically meant bringing in the military. The 1863 Draft Riot ended when federal troops arrived after a long march from Maryland. Detroit’s terrible 1967 tumult ended with tanks on the streets. The National Guard was deployed in Los Angeles in 1992. Trying to stop a riot with too small a force can often lead to more, not less, bloodshed, because as the riot continues, vigilantes step in and beleaguered policemen can resort to brutality.
My colleague Christopher Stone has argued that there is another lesson about fighting riots to be learned from the incidents in the Paris suburbs in 2005, and the violence that didn’t happen during the Republican National Convention in New York City in 2004. In France, the police initially arrested relatively few people, but sought serious criminal penalties for those they did arrest. The New York Police Department arrested more than 1,000 people and let them go. The New York strategy protected the city; the French strategy wasn’t as effective.
The lesson: Light penalties widely applied and serious penalties applied to a few can both deter unlawful behavior. This is a central conclusion of Gary Becker’s path-breaking economic analysis of crime and punishment. But in the case of riots, it is awfully hard to actually prove wrongdoing and extremely important to clear the streets. Arresting widely and temporarily can be far more effective.
Riots strike at the very soul of cities. Even when they are connected to understandable grievances, they do great harm, particularly to the poorest residents. We all want the unrest in the U.K. to end with a minimum of bloodshed and brutality. The best way to achieve that end is with sufficient numbers of police on the streets and gentle but restraining incarceration.
(Edward Glaeser, an economics professor at Harvard University, is a Bloomberg View columnist. He is the author of “Triumph of the City.” The opinions expressed are his own.)
To contact the writer of this article: Edward L. Glaeser at email@example.com.
To contact the editor responsible for this article: Max Berley at firstname.lastname@example.org.