I’ve written a lot in this blog about the precipitous drop in crime over the past two decades. Cause and effect are difficult to pinpoint, largely because each proposed explanation has been mainly tested by itself. Many scholars have looked at the effects of mass incarceration on crime, and many have looked at the effect of the waning crack epidemic. But these and other explanations are all occurring in the context of a big, diverse nation experiencing regular cultural, political, and economic changes.

One potentially fruitful approach to studying why crime has fallen is to compare the experiences of different places to see if any trends emerge. Such comparisons are frowned upon among criminologists. The professional body of academic criminologists, the American Society of Criminology, has issued only two policy proclamations, and one is thou shalt not compare crime data across places. Their reasoning is generally sound: crime data is too easy to fake and thus is relatively easy to move around in the rankings.

To respect that proclamation, I focus on the one type of crime that is really hard to fake: homicide. The interactive graphic below describes the homicide rates of the 24 highest populated American cities. The data are pulled from three periods. First, I rank cities by their homicide rate in whatever year homicide peaked (generally around 1991). Then I re-rank the cities in 2001 and again in 2010. This general strategy is not new. University of Chicago economist and Freakonomics author Steve Levitt used this approach to compare peak homicide with 2001 data. He concluded that the homicide decline was so consistent across cities that there was no need to think about local explanations.

As Levitt notes, from 1990 to 2000 there is little change in where cities rank on homicide.

But there is a lot of change by 2010, suggesting that we should pay careful attention to individual cities’ experiences.

Several cities moved down the rankings at least 3 places. Dallas and New York City dropped 9 places and neither is in the top 10 in 2010. In a list of 24, that is an enormous change. Los Angeles and Houston moved down 6 places, and San Antonio, Denver, and Jacksonville moved down 3. Washington, DC, which was the murder capital in 1990, is now 3rd, and if current trends hold, will move to the bottom of the top 10 in 2012.

On the other side, several cities are going in the wrong direction. Las Vegas stands out as the only city on the list where the murder rate actually increased between 1990 and 2010, moving the city from 20th place to 5th. Though murder rates are down in Indianapolis, the city has moved way up the rankings, from 17th to 9th. Columbus and Milwaukee both moved up 5 places and Philadelphia moved up 4, a dubious accomplishment as the city had already started in 8th place. Baltimore moved from 4th place to 2nd place and Detroit is now number 1 on the list.

So the question is, what can we learn from the changes in these rankings? If you drop Las Vegas from the rankings on the grounds that the Vegas experience is hard to generalize in any context, a clear pattern emerges. If you look at the Forbes ranking of the 100 fastest growing cities, Dallas is in 2nd place, Houston is 4th, and San Antonio is 9th—all cities that saw steep drops in their murder rankings. Cities at the bottom of our list, where homicide was relatively rare and stayed rare are also in the top 10 in growth: Austin, San Jose, and San Francisco.

So, easy story right? Economic growth = crime reduction.

Not so fast.

The Brookings Institution’s 2012 global MetroMonitor ranking of cities, employment, and output ranks Baltimore 4th and Detroit 6th among U.S. cities, way ahead of, for example, New York City, which has a lower homicide rate. And Milwaukee (where the homicide rank is up) is ahead of Denver (where the homicide rate is down).

One possible explanation is that there has not been one, but two crime declines.

Steve Levitt was right—the 1990s were dominated by trends that affected everybody. The crime decline of the last few years, however, seems to be about the unique experiences of different cities. I will offer some ideas in my next blog post about what is driving those city-specific trends, but in the interim, the evidence strongly suggests that thinking about why cities are different in their crime decline experiences is an excellent place to start looking for explanations.