Do apartments really mean higher police costs?
Most of my work in city planning revolves around constructing different types of cost-benefit analysis for the sake of building better cities for people. While that often means I’m putting dollar signs in front of things, like zoning districts or development patterns, our goal will always remain focused on people. People want to feel safe, and we’re willing to pay for the security. Police and fire protection take up a hefty chunk of most cities’ budgets, so we need to build a strong understanding of that connection. Let me start with a sentiment I hear way too often:
“Apartments (and other high-density housing types) require more police services, because they have higher call volumes and crime rates, therefore they cost more.”
This sentiment depends on these variables assumptions:
Apartments, and other types of high-density housing, will have higher police call volumes.
The cost of police protection depends primarily on call volume.
Apartments, and other higher-density housing types, have higher crime rates and higher call volumes
The cost of police service really depends most on the number of police officers on the force. Paying people, especially in a risky occupation, always comprises the highest cost of that service. Then, as the number of officers grows so does the amount of support staff, along with facility and equipment needs such as patrol vehicles. So, this common sentiment suggests that as our population density rises, and we build higher-density housing like apartments, then so will the scale of our police force—thereby increasing overall costs. Well, as I can’t seem to resist doing… I’ve gathered some data and made some charts.
Governing published a decent report a while back outlining data from the FBI’s Uniform Crime Reporting program. I took cities with a population of 100,000 or greater and matched them up with some population statistics and city size data to arrive at population density numbers. When we bring these different metrics together we get a glimpse of the relationship between the size and density of cities with the size and density of their police force.
Total population and total number of police officers
First, we can easily see a rough relationship between the total population of the city and the total number of police officers:
Please don’t try to read the actual names of the cities, you’ll hurt yourself. Besides, the chart only labeled every third city or so because they couldn’t all fit. Based on this arrangement of data we can observe a trend line sweeping down from the top left and flattening out toward the bottom right. This suggests a correlation between the total number of police officers and the total population of a city. New York City on the far left has by far the largest population and easily the highest total number of police officers.
A scatter plot also shows the relationship well. With a scatter plot you’re looking for a pattern in the dots. Here you see a definite upward trajectory from the bottom left to top right. As you look left to right the population of the city grows, and bottom to top the total number of police officers grows. Again, we see New York City as an extreme, but it still falls in line with the overall trend. Based on this small example we can conclude that some relationship exists between the population of a city and the total number of police officers.
One interesting note on this scatter plot before we move on. Notice the dynamic between New York City, Chicago, and Los Angeles. Los Angeles has approximately 4,000,000 people and 9,850 police officers. Meanwhile, Chicago has approximately 2.7 million people and 11,954 police officers. Los Angeles has 32.5% more people with 17.6% fewer officers.
Population density and officers per 10K people
So, if it’s true that as population density grows and the development patterns which house it become more prevalent that crime and police call volume increase, then the police presence should also rise similarly. These next charts illustrate the relationship between the population density and the average number of officers per 10,000 people (officer density) of those same cities. Remember, the costs of police service has a closer relationship to the number of officers and personnel than call volume. We’re looking for a higher officer per 10k people to reveal to us a rising call volume.
The chart above does not show a correlation between population density and officer density. If higher population densities had a strong correlation with higher crime and call volumes, and therefore rising costs for police protection services, then we should see a rising pattern from bottom left to top right. This chart doesn’t show much of a pattern at all. Next, the scatter plot:
This scatter plot has much less of a discernible pattern than the first one when we compared total population with overall number of police officers. We can find a decent baseline for lowest number of officers per 10k people a city might expect to employ, but the overall pattern doesn’t tell us much. None of the population density segments, moving left to right, have much clustering. While we don’t see much overall pattern from these two charts, we might find some subtler patterns if we incorporate an additional metric.
While I was gathering the population and law enforcement data I came across some common city and neighborhood ranking lists. Neighborhood Scout, a real estate investment analytics groups, publishes a yearly “Top 25 Most Dangerous Neighborhoods” list. I tagged any cities that showed up more than once on their list between 2015 and 2018 with the color red. Then, I also looked over Business Insider’s “The 33 Safest American Cities to Live In” piece (March, 2018) for any cities I had data for and tagged them green. The next two charts add the red and green tags to the previous two charts:
The red and green tagged cities don’t show much correlation across different population densities; however, we can see a correlation to number of officers per 10k people. These red “dangerous” cities have more officers per 10k people than other cities with the same population density, sometimes more than twice as many. Likewise, the green “safe” cities generally have fewer officers per 10k people. In either case density doesn’t seem to relate to the danger or safety level of the city, or the number of officers per 10k people.
New York City, located in the far top right corner, has over 28,000 people per square mile with 42.3 police officers per 10k people. Chicago has a density of 11,900 people per square mile with 43.9 officers per 10k people. That means New York City polices more than twice the population in the same amount of space with 1.5 fewer officers per 10k people. In this case we might expect Chicago to show up on the “dangerous” cities list and have a higher per capita cost of police protection.
The table above describes a sample of the data used in the charts. On the left you can see that our “dangerous” cities have an officers-per-10k-people range of 28-44. That’s altogether higher than our “safe” cities which have a range of 12-20 officers per 10k people. However, they don’t differ much in population densities. In fact, the “safe” cities have a higher average population density at 6,054 people per square mile than the “dangerous” cities at 4,546. That’s a 33% higher average density in our “safe” cities, with 60% fewer average officers per 10k people. That seems to directly contradict the notion that denser cities, with their higher density housing types, have higher crime rates needing more police officers and costing more money.
Interestingly Neighborhood Scout’s 2018 “Most Dangerous Neighborhoods In America” list includes a picture from each neighborhood on the list. Scrolling through these pictures one can see a variety of development patterns, including numerous pictures of single-family homes. While the development types differ, they all appear old, dilapidated. I’d encourage anyone interested to peruse these neighborhoods in Google street-view. You’ll see a variety of development patterns.
I put this together quickly so it’s not a fine-tuned study, and some of the methodology makes my academic side cringe a bit. I can immediately see a variety of directions I could take this study where it could tell a more complete and detailed story. So, this feels a little like sharing a rough draft or incomplete work. However, we’re not currently having a very detailed or complete conversation about this issue as it is.
I’ve seen the “apartments (and other high-density housing types) require more police services, because they have higher call volumes and crime rates, therefore they cost more” argument carry significant weight in public meetings and influence policy decisions without any support data at all. The scary part is that currently it’s not an academic argument. Most times I’ve heard this sentiment expressed by a resident or public agency describing one particularly problematic housing complex, and then generalizing that scenario across every other type of apartment or high density housing up for discussion. It boils down to asserting that “criminals are more likely to live in duplexes, triplexes, quadplexes, apartments, and mobile homes than single family homes so we don’t want those types of housing here,” with zero evidence to support it.
My hope here is to add a little data to an important discussion, and hopefully inspire more robust attempts to study this relationship closer.
The original sentiment about apartments depended on three variables. Let’s revisit those variables:
1. “Apartments, and other types of high-density housing, will have higher police call volumes.”
The 11 highest-density cities in this study all have fewer officers per 10k people than the densest “dangerous” city, Chicago. If the above assumption were true we should see a rise in officer density correlating to a rise in population density. This data does not show that correlation.
2. “The cost of police protection depends primarily on call volume.”
Police department budgets depend primarily on the number of officers, support staff, and equipment/facility needs. Staffing and equipment needs depend first on a level of availability, not call volume. The cost of police service remains mostly constant between a zero call volume and the highest call volume a static number of officers can handle.
3. “Apartments, and other higher density housing types, have higher crime rates and higher call volumes.”
The most “dangerous” and “safest” cities lists used in this study show a variety of housing types. The most dangerous list has neighborhoods with apartments, but also with single family homes. They don’t have a housing type pattern. However, they do have patterns of dilapidated housing and vacancy. Vacancy touches on a population issue, but it’s not too much density; it’s a lack of population relative to the capacity of the built environment. Similarly, the cities on the “safe” cities list show a pattern of structural condition and high occupancy, but not housing type. Those cities have housing stock that’s either new or well taken care of. They also have a high occupancy rate, which usually leads to higher population densities rather than lower. In the above data, density doesn’t seem to relate to the danger or safety level of the city.