Powered by TypePad

« Those Pesky Reporters | Main | Kristof Gets Results... Or Does He? »

June 16, 2005



Are crime rates calculated based on reported crimes, arrests or a combination of the two? If arrests are a component, you would think that more police = more arrests, therefore a higher crime rate. WRT the terror alerts, again, if arrests play are part in the stats, you might suspect the drop in the rate was due to policemen being focused on building or structure security vs. "normal" crime.


Philadelphia had Operation Sunrise a couple of years ago. The goal was to sweep into a neighborhood, clean out the crack houses, have a strong police presence for a time, and then move on to the next neighborhood.

Oddly enough, the people in a neighborhood that was cleaned out protested the absence of police officers who'd moved on to the next neighborhood.

It seems crime went down during the times there was increased police presence on the street and in patrol cars. Of course, most of these neighbors, perhaps all, were not social scientists so they may have mistaken correlation for causation.

But they thought having cops around kept criminals away.


The good thing about more cops is, they're a more direct deterrent. A guy who may not care the difference between a 10- and 20-year sentence because he thinks he won't get caught will still recognize that the cop on the street corner means a greater chance of getting caught.

It's the old formula: motive, means, opportunity. More cops = less opportunity.


I think that the authors here are overstating the extent to which "more cops less crime" was not believed. It is true that it is hard to sort out a model in which cops and crime are exogenous from one another, but this is not the first time that it has been done. Steve Levitt did something similar ages ago to prove that Braxton's "broken window" approach to policing added zero value when one took into account the increase in manpower.

Jon Klick

The data we used was reported daily crimes, as opposed to arrests (we would have liked to examine arrests as well, but, amazingly, the DC police told us they did not have daily arrest counts to give us . . . I suspect if we had pushed a bit more, we might have gotten those numbers). I don't know that we overstated the degree to which people didn't believe more cops => less crime since we basically say in the JLE article that economists have uniformly discounted the possibility that the studies finding no/positive effect of cops on crime are successful in isolating a causal effect (though the criminology and sociology literature is less uniform in this respect). I think our contribution is that we offer a cleaner test than has been used in the past, and it offers more opportunities for replication (i.e., can be done for any city providing daily crime numbers and a description of how it changes its police staffing when the terror alert level changes). Additionally, as my co-author points out in the Times article, it's important not just to figure out the direction of the effect (which is fairly intuitive) but the magnitude as well, given how much we spend on policing and the cost of crime. Thanks for the comments.

The comments to this entry are closed.