University of Wisconsin–Madison
Podcast Icon

Jamein Cunningham on How Segregation Affects Homicide Rates

  • Jamein Cunningham
  • November 21 2022
  • PC121-2022

Jamein Cunningham
Jamein Cunningham

High levels of segregation can have significant impacts on communities and the individuals living in them. New research uses railroad tracks as a measure of segregation and overlays data on homicide deaths to determine if people of color living in highly segregated communities are more at risk. In this episode, Dr. Jamein Cunningham shares the findings in the paper he co-authored with Robynn Cox, Alberto Ortega and Kenneth D. Whaley titled “Black Lives: The High Cost of Segregation.” Cunningham is an assistant professor in the Jeb E. Brooks School of Public Policy at Cornell University and an IRP Emerging Poverty Scholar Fellow.


Siers-Poisson [00:00:05] Hello and thanks for joining us for the Poverty Research and Policy podcast from the Institute for Research on Poverty at the University of Wisconsin, Madison. I’m Judith Siers-Poisson. For this episode, we’re going to be talking with Dr. Jamein Cunningham about the paper he coauthored with Robynn Cox, Alberto Ortega and Kenneth D. Whaley titled “Black Lives: The High Cost of Segregation.” That was published by the Washington Center for Equitable Growth in their Working Paper series. Dr. Cunningham is an assistant professor in the Jeb E. Brooks School of Public Policy at Cornell University with a joint appointment in the Department of Economics. He’s also a faculty affiliate at the Cornell Population Center and the Cornell Center for Social Sciences, and is a current IRP Emerging Poverty Scholar Fellow. His teaching and research interests include urban economics, economics of crime, and economics of race and discrimination. Jamein, thanks so much for joining us today.

Cunningham [00:01:01] Oh, I’m glad to be here.

Siers-Poisson [00:01:03] So how did you go about measuring segregation?

Cunningham [00:01:07] Well, we use the traditional measure that we see in sociology and economics, the Index of Dissimilarity. It has several limitations, but it does allow us to kind of get at least a good degree of unevenness, as far as where different demographic groups lie within a large area. And so the Index of Dissimilarity is a measure that’s going between zero and 100, meaning zero, you have a perfectly integrated society and 100, you have a perfectly segregated society.

Siers-Poisson [00:01:45] And now you used railroad tracks as a way to look at where these groups were living. Why railroad tracks? What is it about that particular part of infrastructure?

Cunningham [00:01:56] Yeah, I mean, you hear the term like someone who lives on the other side of the tracks. And then the question is, you know, is that anecdotal, or is that something that actually exists? You talk to people and they say, yeah, I clearly remember crossing on the other side of the tracks and noticing a big difference between those locations. And where I’m from, there’s an old railroad track that’s kind of covered up because, you know, no one uses it, it’s an old manufacturing, small town where the plants, you know, they closed down in the eighties. And so that track was covered up by grass. But if you kind of looked at where that track went you can kind of say, hey, this is one community, and on the other side it was another community. So I had some kind of notion of like, “what does that mean?” right in my daily life. And so we use the railroad tracks because it provides a nice feature that, you know, has been used to allocate people across geographical boundaries, meaning that these railroad tracks are going to be laid, and at least the majority of them were laid in the late 1800s or early 1900s. And so they’re going to exist prior to these migration patterns that we know. We have 6 million African-Americans leave the South and go to a northern and western MSAs. And these railroad tracks is A, not THE, but it’s a way to kind of allocate people to certain communities. And so we’re able to exploit the fact that these railroads exist. They’re somewhat independent or exogenous to migration patterns, but they can be used as a technology to allocate people across geographical locations.

Siers-Poisson [00:03:28] And I’m going to ask you to define what an MSA is, because I don’t think everyone knows.

Cunningham [00:03:33] Oh, a Metropolitan Statistical Area. That’s what the census use to define these economic zones where economic activity or labor market exists. And so you want to think of like, Detroit would be a principal city, but the MSA will include the surrounding suburbs, where people may commute to or from for work. So you want to think of, a city is going to be a political unit and you want to think of a MSA as an economic unit.

Siers-Poisson [00:04:07] Thank you. And so does using railroad tracks work mostly for urban areas, or can you apply that to other areas as well?

Cunningham [00:04:15] That’s a good question. In our paper, we focused mainly on 120 northern MSAs. We follow previous work by Elizabeth Ananat, who creates this railroad measure that separates or provides some kind of measure of segmentation of communities. So this is not our work, we’re leaning on her previous work. And then also another limitation of our study is that it focused on northern MSAs. And so there’s a question of does this apply to rural locations or does this apply to southern places? And we don’t know from our analysis. One can imagine that even in a rural area where there’s a train track, that there might be some allocating of individuals across these train tracks. But our analysis for these northern MSAs — which vary in size. None of them are rural — does show evidence of a strong linkage between having these railroads that segment these communities, and the Index of Dissimilarity, which is a measure of segregation.

Siers-Poisson [00:05:19] So you used that measure of segregation as an overlay for homicides. How did you do that?

Cunningham [00:05:26] Oh, yeah. So what we did is conducted a two stage least squares analysis where we first isolate the relationship between the railroad index and the Index of Dissimilarity. And then we use that information to parse out the variation in segregation that will exist due to the segmentation of these communities due to the railroads. And so you want to think in a sense, is that these railroads provide an opportunity for us to get a causal measure of segregation that may influence homicide victimization.

Siers-Poisson [00:06:06] So let’s talk about it in really concrete terms. So you’re looking at how segregated an area is, like you said, on a scale from 0 to 100, the closer to 100 being the most segregated. And then you pull homicide data and put those together, right?

Cunningham [00:06:23] So we do three things. So in more concrete terms, the first thing we do is take the segmentation of communities, right? So we have this railroad division index which goes between zero and one, right? One means segmented these communities and zero are the least segmented. And we first see if there’s a strong relationship there. So we compare the Index of Dissimilarity to the segmentation of these communities. So that’s the first thing you do. The second thing is that we get information on homicides from the Vital Statistics Multiple Cause of Death files, which gives us information on the manner of deaths for all individuals within the country. And then we can get the information specifically for homicides by race. And we focus on white versus nonwhite, mainly due to the fact that when we try to get at different ethnicities or different races, that those categories are changing over time. But we at least believe and we show in the paper that it matters to include these groups when thinking about the effects of segregation. And then the second stage is that we use the information that we get from comparing the railroad division index to segregation. We use that information to kind of see how segregation may be influencing homicide. And the information that we get from segregation, at least using this index of dissimilarity, is basically taking information from the census about where individuals are located, what is the population in the given census tract from a particular group relative to the total population of the city. And so we’re trying to capture evenness of individuals within the community and then compare that to what we see with homicide victimization. And what we find out is that those places that have high rates of segregation experience much higher rates of homicide. But for only one group, nonwhite civilians. Right. We are unable to find a relationship between segregation and homicide victimization of white civilians, which actually was not our prior. But the discussion with my other coauthors is that, you know, there may be a spillover effect, right? If you have violence in certain communities that may lead to violence in neighboring communities or whatnot. If segregation is going to be tied to poverty and desperate actions, that might increase the victimization of other groups that may be seen to be doing better, at least so far as their socioeconomic status. But we don’t find evidence of that. Now, one may say, right, segregation within the community is meant to separate these individuals and therefore one could have the prior that increased segregation should lead to a decrease in homicide victimization for whites. And so our lack of evidence does not necessarily say that segregation in and of itself doesn’t have spillover effects. It’s just saying that we are unable, given our research design, to link segregation to white homicide victimization. But the relationship for nonwhite homicides is there, it’s robust and it’s persisting over time. So one of the things that we do is we look at the relationship between segregation and homicide victimization in 1970, 1980, 1990, 2000 and 2010. And it’s important that we look at these different time periods, because from 1960, to basically 1990, 1992, crime is increasing. But from 1992 to 2010, crime is decreasing. So we have this inverted U-shape. And so we are able to say even when crime is increasing, there is a strong relationship. But even when crime is decreasing, there is a strong relationship between segregation and homicide victimization of nonwhite civilians. Now that that relationship weakens over time, meaning it becomes smaller, but it’s still there, it’s still meaningful, and it’s still significant in terms of the number of lives lost due to segregation, a lack of opportunity, and a lack of economic or intergenerational mobility.

Siers-Poisson [00:10:40] Let’s dig in to that a little more, Jamein, because you said right there, you know, lack of economic opportunity and, you know, these communities often having, for instance, a lower tax base because the housing values are depressed because of systemic racism and all those things. I think people could look at this and say, how does segregation increase homicide? What is the mechanism that you’re teasing out of that?

Cunningham [00:11:08] So the first thing is that there’s a circular argument, right? That places that are highly segregated, segregation is going to be tied to poverty it’s tied to inequality it’s going to be tied to not-well funded schools. So and then you may think of, you know, the housing situation and so forth and so on. And so the question is which one is it, right? Is it segregation? Is it a lack of resources? Is it lower quality schools? And so what’s nice about — shouldn’t say kind of what’s nice — but what we are allowed to do by using railroads in a way as something that existed prior to migration is that we’re able to isolate the relationship between segregation and many other outcomes. So we can kind of say, well, holding everything else equal or everything else being the same, right? So education levels or quality of schools, current, you know, property crime or not necessarily property crime, but current levels of income or inequality, we can see those places that have higher degree of segregation, have higher homicide victimization rates. Now, what we try to do is say, okay, what can we say as far as the possible mechanisms? So there’s already research that shows that, you know, the black migration is associated with a series of outcomes. There’s work by Leah Boustan and Standard Court that shows that, you know, the black migration is going to be associated with white flight, it’s going to be associated with less tax revenue s housing values depreciate. There’s going to be more policing, targeted policing, in these communities, a more punitive criminal justice system. And but what we’re saying is that even, you know, even controlling or thinking about migration patterns, where people are moving to or the lack of mobility as it relates to choosing on where they stay, matters. And that further leads to more policing. But what we find is that the two things that we find as it relates to the mechanism is A) segregation is going to be associated with less spending for a variety of public goods, including schooling, including policing, including fire safety. However, what’s interesting about this is that there’s going to be an increase in the share of expenditures devoted towards public safety, devoted towards policing. Meaning they’re taking away funds from other public goods in order to provide public safety. And this speaks to the elevated, you know, levels of violence within which we document in the paper. Also what we do show is that the increase in homicide victimization is really being driven by the city center. Right. So it’s going to be driven by a Detroit and not the surrounding areas or Chicago and not necessarily surrounding areas. And so this speaks to the rising inequalities in these communities where the central city is going to be disproportionately people of color, disproportionately lacking adequate housing, quality schools, but also facing higher levels of violence. And so we at least in the paper argue that the primary mechanism in which we are seeing elevated levels of violence is due to the lack of or provision of public goods including educational resources, including public safety, as we see less spending there.

Siers-Poisson [00:14:49] So you looked at specifically homicide rates. What about other types of violent crime? Do those also follow those segregation patterns?

Cunningham [00:14:58] Yeah. So what we do in the paper is we mainly focus on violent crime, well we mainly focus on homicide. But we do look at violent crime as a total category or as a grouping. And violent crime includes murder, rape, assault and robbery, assault being the main component of it. And we do find elevated levels of violent crime. We don’t necessarily look, we don’t break those down by each individual category. We do pull murder out to kind of see if there’s difference in that result. But we understand what we see with murder is going to be a combination of white and nonwhite victimization, which muddies the water in kind of seeing that there’s racial disparities in violence. And so one of the things that we are doing in this paper is breaking out homicides or murder by racial group, by victim. We do look at arrest, right, as outcomes. And so we do look at changes in arrests for violent crime and property crime, as well as for drug related offenses. And we’re finding lower levels of arrests. And this makes sense considering that there’s less spending in total for public safety. So then the question becomes, okay, if there’s less arrests, right, there’s less resources, what are the things that local municipalities or cities or MSAs, what can they can they do in order to fight, you know, higher violence or homicide victimization? And what we see is, is a more punitive criminal justice system. We find evidence of higher levels of admissions in these places that have higher rates, higher levels of segregation. And we also find an increase in incarceration and imprisonment as these places become more punitive. Now, part of that relationship, as it relates to imprisonment, far as individuals who are convicted of homicide, they’re going to serve a longer sentence. But homicide, what it is difficult to solve, most places solve less than 50% of their homicides. Many places, that number is much lower, lower than that.

Siers-Poisson [00:17:13] You talked about the impact of lower spending for a variety of public services, including policing. And certainly there’s a lot of debate about level and type of policing and the effects it has on people and communities of color, specifically. What do you think your research adds in looking at that issue?

Cunningham [00:17:33] Yeah, I mean, so we try to be cautious in the interpretation of our results, right. In the sense we do kind of say there’s some that we do provide that evidence that the lack of policing may be problematic in resolving this issue. We have a work by Aaron Chalfin and Morgan Williams, and I think Ben Hansen and Emily Weisburst on this paper that kind of has the similar tension between policing and homicide victimization and also other types of crime. And it brings into question, you know, what is the optimal number of police? We try to point to the fact that we don’t want local municipalities to spend even more on policing. Right. So even though that in aggregate they’re spending less, they’re spending a larger share of their budget on public safety, meaning they’re sacrificing other public goods. And so what we call for, or at least recommend, is an investment into the schools, into labor market opportunities, into infrastructure, right, to provide, you know, better opportunities for the individuals in these communities instead of just doubling down on policing. Right. Instead of using the hammer, right, and have more police. We do show that there’s less spending on education. Right. And so there can be room for investments there as well. So anything that can provide or generate opportunities is what we push for, not necessarily more policing. There could be room for a better policing and more policing, but we don’t find that as the main way to kind of deal with higher rates of violence. It is a way, but we prefer or at least recommend by increasing opportunities that promote intergenerational mobility.

Siers-Poisson [00:19:19] So where does this research go from here?

Cunningham [00:19:22] Yeah. So now we’re looking at there’s a couple of things, right? The Index of Dissimilarity is one way to kind of measure segregation and it has limitations. There’s a new method that Trevon Logan has used that kind of use individual level data to kind of see who’s your actual neighbor, which can be a nice way to kind of get and measure segregation and also allows us to measure segregation in mainly more rural communities as well. That’s one approach that we might go into. But I really think what we want to do with the next set of papers is kind of get health outcomes. Homicides is one a way in which segregation can be detrimental to any community, but in particular, in our analysis, the black community and now and minority communities more broadly. And then where we also want to go, we want to extend our analysis outside of the north. Right? So the question becomes how generalizable, is this a northern story? And it could be. Right. There’s many reasons to think that, you know, due to the Great Migration and how both the migrants and those individuals who were native to these designation places responded in a way that may have led to these adverse outcomes that could be different in the South, but we don’t know that to be the case. And so there’s room for exploration there.

Siers-Poisson [00:20:52] Well, Jamein, thanks so much for sharing your work with us. It’s really interesting.

Cunningham [00:20:56] Thank you. I appreciate you taking the time to listen.

Siers-Poisson [00:20:59] Thanks so much to Dr. Jamein Cunningham, assistant professor in the Jeb E. Brooks School of Public Policy at Cornell University, and a current IRP Emerging Poverty Scholar Fellow. He joined us to talk about the paper published by the Washington Center for Equitable Growth in their working paper series that he coauthored. It’s titled “Black Lives: the High Cost of Segregation.” The production of this podcast was supported in part by funding from the U.S. Department of Health and Human Services Office of the Assistant Secretary for Planning and Evaluation. But its contents don’t necessarily represent the opinions or policies of that office and the other agency of the Federal Government or the Institute for Research on Poverty. Music for the episode is by Poet Dog Pondering. Thanks for listening.


Health, Incarceration, Inequality & Mobility, Intergenerational Poverty, Justice System, Neighborhood Effects, Place, Policing, Racial/Ethnic Inequality, Social Determinants of Health


, ,