Which Cities Are Becoming More Violent?

Even though COVID-19 death rates have (at least temporarily) begun to decline, cities now face a threat that may last long after COVID-19: a wave of rising violence. In New York City, for example, shootings have nearly doubled over the last year, and homicides are up by a third. 

This wave is quite recent; in some cities, it seems to have began after the George Floyd protests. In St. Louis, for example, there were 70 homicides for the first five months of the year—exactly the same as in the first five months of 2019. But by the end of August, the number of homicides had more than doubled to 185. Thus, there were 115 homicides in June, July, and August alone. To put the matter another way, St. Louis went from 14 homicides per month (pre-protest) to 38.3 (post-protest).

In the long run, rising crime rates may be more threatening to the 2010s urban revival than pandemics or telecommuting. COVID-19 may be quashed in the next year or two by a vaccine, and emptied office buildings can be turned into condos. But if people feel unsafe walking in cities, those cities are unlikely to be able to compete with their suburbs, and even the people who stay in those cities will drive more and walk less. So if you worry about the adverse environmental consequences of automobile-induced air pollution, you should be worried about this trend. (And because the victims of increased violent crime are disproportionately persons of color, this is a racial justice issue as well.) 

To make matters worse, our most widely quoted politicians don't have coherent answers to the problem. Rep. Alexandria Ocasio-Cortez has said, “Maybe this has to do with the fact that people aren’t paying their rent and are scared to pay their rent and so they go out and they need to feed their child and they don’t have money so … they feel like they either need to shoplift some bread or go hungry.” But in fact, most property crimes in New York have actually decreased over the past year. Robberies are down slightly, grand larceny is down 20 percent, and petty larceny is down 8 percent. Because the crimes that have increased most significantly are not crimes related to need, Rep. Ocasio-Cortez's statement seems to be supported by no evidence whatsoever. President Trump's commentary has also not been helpful; he is understandably focused on attacking Democrats in order to get reelected.

So if need isn't the problem, what is? I'm not sure, but I thought I would dig around city crime statistics to see if I could find any clues. In particular, I wanted to know a) how typical is New York's summer wave of gunfire? and b) if not every city has had rising homicide rates, is there a difference between the cities that have suffered from large homicide increases and those that have not?

Month-by-month homicide data is not available for every city,* so my sample size is a bit small. Nevertheless, I was able to dig up information on about two dozen major central cities. I found that several cities seemed to have no increases at all this summer, and that the rest were split evenly between cities like St. Louis and New York, that had suffered major increases, and cities where there had been a more modest increase.

In four cities, homicides were actually stable or declining this summer. For example, in Houston there were 153 homicides in the first five months of the year (or 31 per month) and only 82 (or 27 per month) in June, July and August. Similarly, in San Antonio there were 60 homicides in the first five months of the year (or 12 per month) and 33 in the three summer months (or 11 per month). In Phoenix, there were 80 homicides in the first five months of the year (or 16 per month) and 49 since June 1 (also 16 per month). San Diego has no statistics for August yet, however, there were only six homicides there in June and July, as opposed to 20 (or four per month) in the first five months of the year. 

In nine cities, there were fewer homicides in the last three months than there were before June 1, but the number of deaths was close enough to the pre-June 1 figure that the number of homicides per month actually increased modestly. For example, in Sacramento there were 27 homicides from January to June 1 (or 5.4 per month) and 21 in June, July and August (or 7 per month). Other cities in this category include Philadelphia (from 155 in first five months to 137 in last three), Los Angeles (115 in first five months, 88 in last three), Baltimore (128 in first five months, 90 in last three), Denver (27 in first five months, 26 in last three), Nashville (36 in first five months, 30 in last three), Cincinnati (33 in first five months, 25 in last three), San Francisco (20 in first five months, 17 in last three), and Seattle (16 in first five months, 13 in last three). 

In eleven of the cities I looked at, the summer got much worse: the number of homicides in the last three months actually exceeded January-May homicides; as a result, homicides per month increased by over 50 percent. These included New York (increasing from 141 to 149), Chicago (261 to 274), Dallas (34 to 40), Boston (16 to 25), Columbus, Ohio (33 to 46), Atlanta (23 to 30), St. Louis (70 to 115), Minneapolis (18 to 20), Las Vegas (36 to 38), Louisville (48 to 54), and Portland (from 5 to 20, the largest percentage increase of the cities I looked at).

Although Republicans often blame social unrest on Democratic mayors, only one of the four cities where murders did not increase (San Diego) is led by a Republican. Although these four cities are a very limited sample, they do have a lot in common. All are Sunbelt cities; all seem to have relatively moderate political cultures. (For example, although San Antonio is a predominantly Democratic city, there are Republican precincts just a few miles from downtown; the same is true for Houston.) By contrast, cities with progressive reputations such as Portland, Minneapolis, and New York had significant murder increases. On the other hand, some other cities with moderate political cultures and significant Republican minorities (such as Las Vegas, Dallas, and Columbus) have also had significant homicide increases. So even though a correlation between political moderation and stable homicide rates seems to exist, I am not sure that it is strong enough to suggest a causal relationship.

Because some commentators blame crime on concentrated poverty, one might expect the most socio-economically segregated cities to have rapidly increasing levels of violence. One way to measure this is the concentrated poverty rate: the percentage of the region's poor who live in a census tract where over 40 percent of residents live below the federal poverty rate. Here, there was no distinction among the three types of cities. In the regions containing four cities with stable homicide rates, the mean concentrated poverty rate was just over 22 percent (ranging from San Diego's 14 percent to Phoenix's 34 percent). The cities with rising homicide rates similarly averaged a 22 percent concentrated poverty rate.**

Given the sheer pointlessness of many recent shootings, it would be noteworthy if there was a correlation between mental illness and rising homicide rates. I do not know of any way to measure mental illness by city. However, it would make intuitive sense that high rates of COVID-19 infections might lead to higher levels of stress. If this was the case, cities that were in counties with high COVID-19 levels would have rising homicide rates. A look at Northeastern cities may support this view; major Northeastern cities have suffered from rising crime rates and unusually high COVID-19 death rates. On the other hand, Houston, San Antonio and Phoenix all have had average-to-high COVID-19 death rates. Harris County (which includes Houston) has 51 COVID-19 deaths per 100,000, higher than nine of the nineteen***counties that include cities with rising murder rates. Bexar County (which includes San Antonio) and Maricopa County (which includes Phoenix) have even higher COVID-19 death rates. Yet Houston, Phoenix, and San Antonio have all had stable homicide rates this summer.

Since the protests arose out of police brutality, it might be the case that the most successful cities have had a different type of quantity of police reforms than other cities. But because there is no way to quantify this sort of thing (or even discuss it intelligently without more detailed knowledge about individual cities) I am not ready to a venture an opinion on this subject.

Another possible explanation is that cities that had longer or more violent protests and riots had more crime.  This theory makes intuitive sense to me, especially since in the late 2010s the cities with the most widely publicized police brutality scandals (St. Louis and Baltimore) also had unusual crime increases.  However, I do not know of any way to quantify riot intensity, so I am not sure this hypothesis can be proven or disproven. 

So what have I learned? First, there is a wide variation in crime patterns among cities over the last several months. In some, homicide rates have dramatically increased; in others, they have barely budged. Second, it appears that the most successful cities are more politically moderate than the rest; however, there are enough exceptions to this generalization that I am not sure that it has any probative value. Third, the most successful cities appear to have about the same amount of concentrated poverty and pandemic deaths as those where homicides have increased. I hope that by the end of the year, crime will go back down to 2019 levels—but if not, we will have more data and maybe we will know more. 

*I got most data from city police web pages and from communitycrimemap.com 

**I note, however, that Brookings data on segregation and concentrated poverty focuses on metropolitan areas rather than central cities; it is not clear whether this should matter. One might also look at racial segregation; however, these cities differ so widely in their ethnic makeup that I am not sure that racial segregation levels are very relevant. Some of the large cities discussed above are majority Black, while others (such as San Diego, Phoenix and San Antonio) are heavily White and Hispanic but less than 10 percent Black.

***Not counting New York City, which consists of five counties.