Podcast
Important Information:
The Economics of Crime and Policing
Important Information:
Ray Owens and Santiago Pinto discuss current trends in crime rates, the challenges of determining the extent of crime and its causes, and the impact of policing on crime rates. They also review research on racial bias in policing and the impact of police violence against Black citizens. Owens and Pinto are senior economists at the Federal Reserve Bank of Richmond.
Speakers
Related Links
Transcript
Jessie Romero: Hi, I'm Jessie Romero, director of research publications at the Richmond Fed. Thanks for listening to "Speaking of the Economy." You can find us and subscribe via your favorite podcasting platform. And if you like what you hear, please tell your friends and leave a review.
I'm excited to be talking today with Santiago Pinto and Ray Owens, senior economists in the Richmond Fed's Research Department. We'll be discussing crime trends and policing, nationally and here in the Richmond Fed's District, which Santiago and Ray have written about recently in our magazine, Econ Focus, and Economic Brief series. We'll post links to those articles on our show page for folks who want to check them out.
Santiago and Ray, thanks so much for being here.
Ray Owens: Our pleasure.
Santiago Pinto: Hi, Jessie.
Romero: As you note in your Econ Focus article, crime is really hard to measure. Actually, there's a few different sources for data and they're all a little bit imperfect in their own ways. But with those kind of caveats, what is the general trend in crime over the past few decades?
Pinto: If you look at the crime data, in general, you will see a relatively sharp decline from 1990 to 1999 in both violent crimes and property crimes. But then that will be followed by a gradual decline over the next two decades. If you look more recently at the UCR [Uniform Crime Reporting] data, which is the data that is most commonly used to track crime rates, property crime has declined. But the violent crime rate has increased in the last year, or last year we have data.
Romero: Are there other differences in crime rates, either by demographics or by location?
Owens: Typically, we think of crime rates as being higher in urban areas, and often that's been the case. But if you look at crime rates more recently — especially for violent crimes — there's some evidence that suggests that violent crime rates are rising, regardless of whether the area is urban or suburban or rural. We're seeing at least those sync up in a way; it varies over time, by type of crime and by the endpoints that you measure by.
Romero: Are there theories about why violent crime might have increased recently, even while property crime has declined?
Pinto: If you look at violent crime, definitely one of the main components that people really track closely is the murder rate. It's not uncommon for that category of crime to show unexpected fluctuations year to year. It's definitely true that we have observed in the last year, from 2019 to 2020, a spike in the murder crime rate.
The issue there is that it's happening everywhere. It's happening all across the board. You see that in urban areas, suburbs. You see that in rural areas, Democrat-run cities, Republican-run cities. You see it all across the board. So, if you want to come up with an explanation, it has to be an explanation that covers all of these possibilities.
There are a lot of explanations out there and we don't really know why that happened. In fact, we can think of a mix or a combination of different factors that have played a role in explaining this spike in the murder rate. People out there are considering three basic, broad theories. But again, we have to be humble in terms of understanding that there are micro mechanisms that drive this spike in the murder rate. It's not completely clear.
People usually try to explain the spike in the murder rate by looking at the pandemic. As we know, COVID disrupted a lot of the social services, even closed schools. But there is perhaps some weakness in that theory in terms of the timing because the spike in crime came a lot after the beginning of COVID.
Romero: Okay.
Pinto: Also, if there was a pandemic specific factor, then we would have seen this all around the world. And you don't see this happening around the world.
Other people out there in the research environment, they also attribute this as changes in policing. We know that there has been a lot of demonstrations involving racial justice and riots. There is evidence that some police may have pulled back on corrective, anti-violence policies. And there's this impact on the people's perception or confidence on police action, so that might have been affecting some of the outcomes.
There [are] also some people that say there are some more guns out there. Purchases of guns have increased in the last few years. And, even though during the pandemic we observed [fewer] arrests, a larger proportion of those arrests were found with people carrying a gun. A lot of people also suggest that a high percentage of the murders involve the use of guns.
So, we don't have a definite an explanation of why the murder rate has increased. For sure, a combination of different complicating factors may lead to experiencing such behavior.
Romero: Yeah, kind of a perfect storm of things coming together.
Pinto: Exactly.
Romero: One of the things that I found really interesting reading the Econ Focus article is that people don't tend to have very accurate perceptions of crime rates, right? You noted that people think crime is going down in my city but up in the U.S., when actually maybe that the opposite is true. I was wondering if you have any theories about why that is. Why don't we perceive it accurately?
Owens: One thing about perceptions of crime — they seem to change over time among different groups in the U.S. We think of this as maybe being a result, in part, of different kinds of exposure to and types of media coverage.
For example, in a Gallup poll released in late 2020, there was a divergence in the percentage of people who thought crime was rising in the U.S. compared to crime rising in their local neighborhood. That gap rose to 40 percent, which was by far the highest in the nearly three decades that the Gallup poll has been measuring this kind of thing.
They speculated that there might have been several reasons. One is pandemic related. Their thought was, with more people at home and not traveling around as much, people probably noticed fewer strangers, outsiders in their neighborhood. That may have constrained somewhat the perception of crime in their neighborhoods. But also, with people at home more, they were probably exposed to the media, television [and] the Internet more.
We've seen a divergence over time in media outlets. They've become more polarized in how they report certain events, to take the spin that they put on it. The Gallup poll tends to find this seems to be a factor that leads to people reporting these national differences. They can't observe them directly, so they have to get them through some other channel and they suggest that, perhaps, this changing media landscape is partly responsible.
Pinto: Let me also emphasize the importance of perceptions of crime. We know what drives changes here are the perceptions — whether you would allow your kids to play outside, whether you would like to walk or jog or go around the city. That's a relevant factor. But, as Ray was pointing out, it is important to understand how people respond to the Gallup question, you need to understand how people make those perceptions, those beliefs. One way has to do with the access to information.
Owens: Real quickly, just the last thing I'll note on this. The Gallup poll found that while people thought crime was rising nationally, they were as comfortable going outside as they had been before. This idea that their locality was more dangerous certainly wasn't affecting that kind of behavior.
Finally, I guess it's worth noting that while they thought crime was rising more nationally, they didn't think the severity of the crime was getting worse, which was sort of an interesting sideline.
Romero: That makes sense. I can see that, you know, I'm sitting at home. My suburb seems very safe when I look outside, but then you watch the news and it's like, "Oh, my gosh."
What do crime trends look like in the Fifth District?
Pinto: Within the Fifth District the UCR data, the data that I referred to before, also shows that the crime rates have followed the general declining trend that we mentioned before that was observed at the national level.
But there are some discrepancies between the states within the Fifth District. For instance, property crimes and violent crimes in the District of Columbia are the highest, not only within the Fifth District but also nationwide. In South Carolina, crime rates are generally above the national trend. The violent crimes in Maryland are about the same as the ones in South Carolina and property crimes have been following the national trend closely since at least 2005. On the other hand, Virginia crime rates are lower than the national average. Violent crimes in North Carolina seem to track pretty well the national [trend].
The other thing is to focus on cities within the states because they showed a somewhat different picture. For instance, in a big city like Baltimore, data that we've been using from the Open Baltimore initiative show that while property crimes follow the declining trend that we mentioned before since 2011, violent crimes have increased.
Romero: So, crime obviously leads to a whole host of negative effects, right? There's the harm to the victims. But then, as you note in your Economic Briefs, there [are] effects on neighborhood growth. It can lead to increased racial and income segregation, lower housing prices. Businesses make their location decisions based on crime rates.
Given all that, of course, it makes sense that policymakers are very interested in reducing or eliminating crime. There are lots of essential strategies to do this. The one that you look at specifically in your Economic Brief is the effect of adding more police. This seems to me, or maybe to the average person, it would be an easy thing to measure, right? You look at how many police there are, you look at how many crimes there are, and then you say one effected the other. But it's a lot more complicated than that. Can you tell us why that is?
Pinto: That's correct, Jessie. That is what keeps us engaged in this literature, because we are trying to quantify and identify the causal effect of police on crime. It sounds easy, but it's really very, very complicated for a number of reasons, some of those you've said already.
One thing has to do with how you measure crime. There are different ways of measuring crime. We have the data from the FBI. We have data from the National Crime Victimization Survey. There are different sources of data. But they are all proxies. It's hard to find out exactly what the true level of crime is.
Not only that, it's also hard to measure the number of police out there. You can think about how many people are there, how many police officers are there. But then there's this question about the intensity [of policing] — how much effort you put in — and that's hard to quantify. There [are] also decisions about, for instance, would you consider the administrative parts in a police agency part of the police force or not. They may be important, right, because these are the people that are actually recording the crime. Some police agencies report that they don't have enough staff to record the offenses that have been reported. That has an impact on people's willingness for them to report crime.
Another thing — and this is one of the most important things — is that police intensity and police presence is not completely exogenous. It's not that you put a set amount of police and see what happens to crime. What happens, in most situations, if there is more crime they you will put more police, right? So, you have this problem in identifying the cause and effect. What is driving what? In the old times, you used to have some studies that didn't [take] that into account and then they were finding a positive effect. It's kind of a weird thing that police is actually increasing crime when maybe it is the opposite saying, okay, since there is crime in this place let's put more police. That's something that has to be considered and taken into account, and it's not really easy.
There's another issue that has to do with the role of policing and the deterrence effects and the incarceration effects. The police may have a role more in a deterrence effect than an incarceration effect. But at the same time, if the police [are] out there and catch a person committing a crime, then he will go to jail, right? Then, depending on what are the rules and what is the time that this person has to spend in jail and so on, it may affect the crime level. But that's not attributed to police efforts, it's attributed to something else. So, it's important to be able to disentangle the two effects of the deterrence effect of police and the incarceration effect, which has to do with the institutional structure, the legal and judicial system.
Another important factor that people have looked into is the probability of arrest. If you think about criminal actions as being the outcome of a rational decision, then it's important to understand what is the probability that you are captured if you commit a crime. Is that related to the amount of policemen or can it be related to the visibility of policemen? You can have similar number of police but they are more visible than others and this has a different effect, so you need to take that into account. That may affect the calculation that potential criminals have in their minds and may affect the level of crime that you observe.
There is this issue about the reporting bias. The UCR data or the FBI data relies on police agency data, and [in order for] the police to record a crime you have to have someone that goes to the police station and report a crime. But there are a lot of other crimes that end up not being reported. The National Crime Victimization Survey reports that 40 percent of the incidents are not recorded. We may need to take into account that police actions may affect incentives for people to have reported a crime, or the size of the police force may affect people's willingness to go to the police station and report a crime. That's something that may bias the results when you're trying to establish that causal effect.
Romero: I guess I just want to try to clarify in my own head. Even if you have tons of police out on the streets and you don't have the administrative force on the backend that can process all those arrests or do anything about it, people say, hey, even though there's all these police out here, why am I going to bother to report anything because I know nothing is going to happen?
Owens: I think that's right. I think Santiago is describing a system we have that would include policemen on the streets, prosecutors and defenders in court, all through, as you mentioned, incarceration. Each of these players are anticipating the actions of other people in this chain. And over time, that can change and it can change their incentives and how they deal with it. This, of course, as Santiago is pointing out, flows down to the level of crime committed at the street level. It's a very dynamic process and far more complicated than I guess we often casually view it to be.
Romero: Yes, definitely.
Pinto: The final issue that I wanted to point out regarding the measurement, the quantification of the impact of police on crime has to do with the area that you are considering in your analysis. You might say, okay, I put more policemen in this area and I see a reduction of crime in this area, but you might see an increase of crime in the neighborhood.
Owens: The police had to come from somewhere.
Pinto: So, in a way, what you see is that police actually displace crime. It's not that you see a reduction in crime, you just displaced it. Depending on which area you are considering, if you're considering the narrow area where you put more police, you will see less crime. But if you extend your area of analysis, you will see that that didn't happen.
Romero: You make a similar point when you talk about place-based initiatives. You look at somewhere that has a high poverty rate and you make a lot of effort to reduce the poverty rate. But really what you've done is you just moved all the poor people somewhere else, so the poverty rate looks a lot lower.
Pinto: That's a great analogy.
Owens: It is the common gentrification argument.
Romero: Given all the complications you just described — I almost feel like I can't ask this — but what does the literature tend to say about the effects of policing on crime?
Pinto: There is some common ground here. After making a huge effort to try to control for all these issues that I was mentioning before, the economics literature supports the view that, in general, a larger police force tends to [cause a] decline in certain types of crime. It has an impact — in some cases, it's larger for violent crimes and for property crimes.
The literature also suggests that the deterrent effect of police, as I was saying before, is more important than the incarceration effect. There [are] also some studies that find that if you invest more in police, if you increase the police force, that may actually reduce the amount of incarceration. To the extent that you think that policy is effective at reducing crime, then you will see less people going to jail. And some research has gone in that direction.
With respect to spatial patterns of crime and location, there is this idea that certain types of crime are concentrating in very small areas in the cities, within a city that are called hotspots. There is quite [a bit of] evidence that this is happening. This is very relevant for police tactics. It's a good way, a measure of how to spatially allocate police resources — targeting police resources to those high criminal, high crime areas seem to have an effect. And, there's not much displacement of crime.
Romero: That reminds me, I wrote an article for the magazine years ago about High Point. N.C., and what they had — this open air drug market that was this plague on the community. They were able to watch and they said, these are the 10 most active drug dealers in this place and we're just going concentrate all our effort on them. Once they got rid of them, the open air drug market just went away and it wasn't replaced. They just focused exactly on that.
Pinto: As much as I highlighted the impact of this potential displacement of crime, for certain kinds of crimes they have to be in a certain area.
Romero: In recent years, there have been some very high profile and tragic examples of police violence and police bias. Does the research find a link between how police behave and a suspect's race?
Owens: In short, yes. I'll let Santiago describe a number of these papers.
There does appear to be, in general, a racial component of police actions. For example, there are papers that find that white officers tend to stop non-whites at a greater rate. But conversely, the same paper also finds that Black officers tend to stop whites to a greater degree than Blacks. So it's not completely asymmetric. And, in that most police departments across the country are composed primarily of white officers, this kind of behavior could then result in a greater number of incidents against Black citizens.
Pinto: This is a very complex issue. It's not only a very delicate matter, but from an academic and technical standpoint it's very complicated to able to disentangle the effects that are taking place in these interactions between police and the community. It is important to understand that there are some limitations in terms of how is it that we can address this.
There are some very recent papers that have directly addressed this with very interesting data. As Ray was pointing out, the evidence seems to suggest that there is some level of racial bias in policing.
Let me give you some examples. For instance, there is some evidence that suggests that the police outcomes are affected by the police officers' race. There is a relatively older study that [found] that adding one more white police [officer] increases the number of arrests of non-whites. But they also show that sometimes the converse is also true — that an increasing number of minority police is associated with a significant increase in arrests of whites.
There is also some evidence using some very new data that suggests that there's racial bias in the likelihood of being stopped by the police and the likelihood of being searched once stopped. Also, there is some racial bias in police investigations. For instance, there's an interesting study that finds that Black drivers are less likely to be stopped after sunset. The reason is what they call a "veil of darkness," which says that after the sun goes down, that makes race substantially harder to discern, what the police can see. So that's one study. There's another one that look at search behavior by police officers and they find that there's a lower bar for certain Black and Hispanic drivers compared to white drivers. Involving police investigations, there is also some research that suggests that officers cite drivers of other races more frequently.
There is also some other racial bias in behavior in terms of the use of force. There is some research that white officers tend to use force more often than Black officers on average and use gun force more than twice as often when responding to similar calls. They also find that when white officers are dispatched to minority neighborhoods, they tend to use any type of force more frequently than minority officers.
Romero: Ray, a few minutes ago you talked about how policing [can be thought of] as a system. It's not just the police force. Of course, the people who live in a community are part of that system. One of the things you mentioned in your Economic Briefs is that the police in the community are complementary. The information has to go both ways. This cooperation can break down when there is a high-profile tragedy and you have some ongoing research on the effect of Freddie Gray's death in 2015 on Baltimore. What are you learning from that research?
Owens: Well, you touched on earlier in one of your questions this core idea about how do you reduce crime. The answer in the literature is often more police, less crime. Santiago clarified that you also have to account for police effort. That turns out to be key in an episode like the aftermath of a tragic incident like that of Freddie Gray in Baltimore in 2015, or more recently George Floyd.
The way we tend to think of this is the police, ideally, exists to make the citizens, the residents of neighborhoods better off by reducing crime, which is very, very costly to the residents. But we know that tensions can build up because of police activity that's too aggressive. It can come to a breaking point, as we sometimes see in the aftermath of one of these incidents.
But what we find is that things change after a Freddie Gray or George Floyd. There is evidence that the local residents tend to report crime less after one of these. Perhaps they don't trust the police as much as they did before. Perhaps, as Santiago touched on earlier, it's more expensive for them in terms of their reputation to interact with the police who now have this tarnished image. But they report less and that provides less information to the police, so they have less information about crimes. So, even if police behavior didn't change at all, the probability of solving crimes would lessen because the police, let's say in this instance are making the same effort, but they have less information.
But, of course, we find that after these episodes, typically police behavior changes as well. The police are more heavily scrutinized. They are well-known articles on police noticing less crime in Baltimore, dramatically less crime, post Freddie Gray, perhaps because they are under heavier scrutiny.
So, you combine a greater reluctance to report crime [and] provide less information with, perhaps, a greater reluctance to follow through on that to the same extent and we can see that, quickly, the beneficial effects of a police force on residents of neighborhoods can deteriorate quite markedly. There is some evidence that that's occurred not only in the instance of Freddie Gray in Baltimore, but also as we saw much more dramatically in the aftermath of a George Floyd-type incident.
Pinto: We know we want interactions between police and citizens to be as complimentary as possible, reinforce each other. When incidents like this happen, that string of confidence is broken.
One of the things that we want to do and understand is if those high-profile incidents affect people's incentives to report cases and to collaborate and cooperate with the police. We will have some preliminary evidence that goes in that direction. Especially for predominantly minority neighborhoods, we see that declining. One way of recording crime is expressed as the number of 911 calls of service. There is some indication that these high-profile incidents, they break this necessary relationship and may trigger something bad. So, we think that this is something important to understand.
Romero: One thing I just wanted to clarify. Santiago, you just mentioned that you saw the decrease in 911 calls after Freddie Gray. Ray, you mentioned George Floyd's death had a large effect, too. It had a large effect on Baltimore, not just in Minneapolis. Do you see tragedies in other cities and the effects carry over to other cities?
Owens: In some cases, especially in the George Floyd incident. It was more localized back in 2015 with Freddie Gray. There were reactions by the minority neighborhoods where the incident occurred there. There [were] also some protests with a bit of violence after the trials of the policeman involved, but it didn't really spill over to any large extent into cities across the country.
Of course, in stark contrast, [the death of] George Floyd did, perhaps, [have spillover effects]. In the minds of many, George Floyd represented that kind of episode, that kind of behavior by police that has been viewed over a long period of time by many people outside of Minneapolis and it apparently resonated with them. Earlier episodes have been notable but tended not to have the broad reaction that George Floyd did.
Romero: All right. Well, Ray and Santiago, thank you so much for joining us and I look forward to reading your further research on this topic.