One Possible Cure for the Common Criminal

The New York Times, "Economic Scene" , June 16, 2005

When Jonathan M. Klick worked in Washington, he noticed a striking effect every time the terrorism alert level went from its usual yellow ("elevated") to the more urgent orange ("high").

"When the terror alert level went up," he recalled in an interview, "you all of a sudden saw zillions of cops around the Capitol and around the Mall."

The pattern gave Professor Klick, now a professor of law and economics at Florida State University, an idea for how to examine a tough social science question: Do more police officers in fact reduce crime?

The answer may seem obvious, but many social scientists have argued that the number of police officers has no effect on crime rates and may even increase them. "If you look at the studies, particularly in the criminology literature, it's either no effect or actually a positive effect," Professor Klick said.

Cities with more police officers have more crime. That is probably because cities with high crime rates hire more officers. But it is hard to separate cause and effect, and, assuming that the officers do deter crime, to figure out how big the effect is.

"We spend a huge amount on police," Alexander Tabarrok, an economist at George Mason University, said in an interview. "So we want to know not simply do police reduce crime, but by how much. Should we have more police?"

He and Professor Klick examine the question in a study published in the April 2005 Journal of Law and Economics, "Using Terror Alert Levels to Estimate the Effect of Police on Crime." (A copy of the article is available here.)

To separate cause and effect, researchers need a "natural experiment" -- in this case, an event that changes the number of police officers for reasons having nothing to do with the crime rate. The crime rates before and after the change can then be compared.

Changes in the terror alert level provided just the sort of natural experiment Professors Klick and Tabarrok needed, because the shifts in police deployment are big, making effects easier to spot. The alert levels -- and hence the number of officers on the street -- go up and down over time, providing multiple tests. And since the number of police officers fluctuates over days or weeks, rather than months or years, any new officers are unlikely to be there because of crime-related expansions of the force.

The two economists looked at daily crime statistics in Washington from March 12, 2002, to July 30, 2003. During that time, the terror alert level rose and fell four times. "On high-alert days," they wrote, "total crimes decrease by an average of seven crimes per day, or approximately 6.6 percent."

Although the Police Department would not officially say how it responds to heightened alerts, the researchers were told unofficially that officers spend an extra four hours on duty after their regular eight-hour shifts. Patrol officers who have finished their shifts elsewhere move to District 1, the area that encompasses the Capitol and National Mall and includes the Smithsonian Institution, the Washington Monument and the Lincoln Memorial.

Not surprisingly, then, much of the drop in crime occurred in District 1. During high alerts, the number of crimes per day in that area fell by 2.62, or about 15 percent. "Almost one-half (43.6 percent) of the total crime decline during high-alert periods is concentrated in District 1," the economists wrote.

To make sure tourists were not just avoiding Washington during high alerts, the economists checked midday subway ridership and hotel vacancy levels, finding no significant difference. Nor did criminals appear to shift from District 1 to other parts of town. Crime was down throughout the city.

A bigger police presence does affect some kinds of crimes more than others. The number of murders, for instance, does not change.

"If you think about what crimes you most expect to be affected by putting more police on the streets, well, it's street crimes," Professor Tabarrok said in the interview. "Theft from automobiles and automobile theft are the classic street crimes, and we found that they fell by a whopping 40 percent during these high-alert periods." Burglaries were also down, by 15 percent.

Since the terror-alert system operates nationally, this research can be replicated in any other city willing to share its daily crime statistics.

But so far, the case for adding more police officers is strong. Using generally accepted cost estimates, Professor Tabarrok said, every $1 to add officers would reduce the costs of crime by $4. The authors did not identify a point of diminishing returns.

"We estimate that if we had a 10 percent increase in police, crime would go down by about 4 percent," he said, adding that researchers taking other approaches have come up with similar numbers. Nationally, he said, "that means about 700,000 fewer property crimes and 213,000 fewer violent crimes."

As a back-of-the-envelope calculation, Professor Klick offered an even more striking suggestion. "It wouldn't be unreasonable," he said, "based on our estimates and based on conservative estimates of the costs of crime, to say it would be cost-effective to actually double the number of people working in police forces, which is pretty amazing."