On this episode of the Matrix Podcast, Julia Sizek spoke with two UC Berkeley scholars whose work focuses on explaining how mass incarceration has changed over the last 30 years.
Alex Roehrkasse is an Assistant Professor of Sociology and Criminology at Butler University. He studies the production of racial, class, and gender inequality in the United States through violence and social control. He was previously a postdoctoral associate in the Department of Sociology at Duke University and at the National Data Archive on Child Abuse and Neglect at Cornell University.
Christopher Muller is Associate Professor of Sociology at the University of California, Berkeley. He studies the political economy of incarceration in the United States from Reconstruction to the present. He is particularly interested in how agricultural labor markets, migration, and struggles over land and labor have affected incarceration and racial and class inequality in incarceration. His work has been published in journals such as the American Journal of Sociology, Demography, Social Forces, and Science.
Excerpts from the interview are included below (edited for length and content).
Q: Let’s start by talking about the main topic at the center of your collaborative research, which is how mass incarceration has changed over the last 30 years. What motivated you to take on this topic?
Muller: It’s useful to step back and try to define mass incarceration. There isn’t complete agreement about how to define mass incarceration, but I think the most influential definition comes from the sociologist David Garland, who argues that mass incarceration is defined by two main features. The first is a scale of incarceration that’s unusual in both historical and comparative terms. This fits the US case because its incarceration rate is so extreme, both in comparison to similar countries and in comparison to its past. From 1970 to 2010, the US imprisonment rate rose from roughly 100 per 100,000 people to roughly 500 per 100,000 people. If you count people in jails, that number gets even higher, to about 700 per 100,000 people. That makes the US a vast outlier with respect to comparable countries.
The second feature of mass incarceration that Garland focuses on is what he calls the social concentration of incarceration. In the US, what he’s referring to is mainly the incarceration rate of young Black men. If you look at the most recent estimates, roughly a quarter of Black men can expect to be imprisoned at some point in their lives. When you zoom in to look at Black men who dropped out of high school, that number jumps to over two-thirds. These are really astonishing numbers, and are part of what has inspired people to try to understand how we got here over time.
One of the main motivations of this project with Alex has been the emergence of a recent debate around this last point – about the relationship between racial inequality and incarceration on the one hand, and mass incarceration on the other. On the one side of the debate, we have a book like The New Jim Crow: Mass Incarceration in the Age of Colorblindness by Michelle Alexander. This is probably the most widely read book on mass incarceration, and it focuses mainly on its disproportionate impact on Black Americans, due in part to the War on Drugs, and due in part to the concentration of police in poor, predominantly Black neighborhoods.
On the other side, you have scholars like James Forman, Jr. and Marie Gottschalk, who are sympathetic to Alexander’s account, but who argue that it’s incomplete. In particular, they focus on the fact that mass incarceration has negatively affected many groups beyond just Black Americans, and that it’s particularly concentrated among the poor.
My read of the debate is that it’s been quite civil and collegial. But as it has spun out into wider public arenas, it’s gotten more heated. As I’ve encountered this debate, I’ve had a sense that people have been talking past each other. And so one of the main goals for me in working on this project with Alex was to try to establish a more comprehensive and up-to-date empirical foundation for the debate. I had a hunch that this foundation would help us to see why both positions actually look quite reasonable depending on how you look at the question — depending on whether you’re looking at the direct experience of incarceration, or whether you’re looking at its indirect effects.
What we tried to do in the project was two main things. The first thing was to update previous estimates of racial and class inequality in prison admissions. They hadn’t been calculated since 2002. You would think this would be a relatively straightforward thing to do, but as I’m sure we’ll discuss, there are all kinds of complicated issues related to how you actually estimate these quantities. One of the main reasons we wanted to do this was based on research that’s come out in recent years showing that there’s been a huge shift in the fortunes of people without a college degree. One of the most famous examples of this is the work of the economists Anne Case and Angus Deaton, who’ve shown that there’s been a marked rise in the mortality rates particularly of White people without a bachelor’s degree. We had a hunch that this shift might also be visible in prison admissions.
The second thing we wanted to do in the paper was to look beyond the direct experience of incarceration and look at the indirect experience. This includes looking at people’s likelihood of having a family member imprisoned, and looking at people’s likelihood of living in a neighborhood with a high imprisonment rate. The reason we wanted to do this was because of a whole body of sociological research that has shown how, because of Black-White wealth gaps, for example, middle-class Black people are much more likely than middle-class White people to be the offshoots from poor family trees. That means they’re much more likely to have family members who are poor than similar White people.
We were also inspired by a lot of research, much of it coming out of sociology, showing how segregation has meant that middle-class Black people are more likely than middle-class White people to live in poor neighborhoods. If you think incarceration and poverty are becoming increasingly associated over time, these dynamics are going to influence differences in the relative direct and indirect experiences of incarceration.
Together, we thought these facts suggested that it was possible that racial and class inequality in people’s risk of having a family member imprisoned — and racial and class inequality in their risk of living in a high imprisonment neighborhood — could seriously differ from racial and class inequality in their risk of being imprisoned themselves.
Q: That points to the two challenges of studying mass incarceration: the question of the class and race factors that make one more at risk of being in prison, and the question of the people who are in direct or indirect contact with the prison system. What were your findings when you put these two different parts of mass incarceration together?
Roehrkasse: Corresponding to these two parts that you’re describing, we really have two main sets of findings. The first is that we show that there have been really significant shifts in the contours of inequality in prison admissions in the 21st century. On the one hand, Black-White disparities have pretty meaningfully declined since the late 20th century. For example, at peak levels of racial inequality in the early 1990s, Black people were somewhere between six and eight times more likely to enter prison than similarly educated White people. That’s just an astonishing level of inequality.
To be frank, you don’t often see racial disparities that large in social science. This is not reducible to any underlying educational differences, because we’re comparing like to like here. By 2015, though, the Black-White ratio of prison admissions had fallen to something more like two or three. That’s a pretty significant decline, but it’s important to say that’s still a really big disparity.
On the other hand, inequality between people who had attained different levels of education skyrocketed over the same period. So again, in the early 1990s, people who hadn’t attended college were roughly five to six times more likely to go to prison than people who had attended college. But by 2015, when our analysis ends, people without college were 20 to 25 times more likely to go to prison than people who had been to college before.
Our second set of findings adds some nuance to this picture. In two separate analyses, we examined people’s likelihood of having a family member in prison, or of living in a neighborhood where a high proportion of residents in that neighborhood go to prison. In both of these cases, we find that Black people with the highest levels of education or income are actually more likely to experience indirect contact with the prison system than White people with the lowest levels of education, or the lowest levels of income.
Ultimately, what we find is that while class inequality in prison admissions now appears to dominate racial inequality, it’s racial inequality that still predominates in other aspects of the lived experience of mass incarceration. Depending on whether we look at these direct or indirect experiences of the prison system, we’ll come to different conclusions about whether race or class matters more. Rather than trying to decide which is absolutely more important, we’ve become much more interested in trying to understand how racial and class inequality interact, and even how these interactions could create opportunities for new alliances to combat mass incarceration.
Q: Can you talk more about how you decided to use education as a proxy for socioeconomic class status?
Muller: The main reason is just data limitations. When people are admitted to prison, they’re not asked about their income, and so we’re forced to use their level of education. We use education as a proxy for class. This is clearly an imperfect measure, and there are all kinds of quibbles you could have with it. But on the other hand, the work of Case and Deaton shows that having a college education is an increasingly important determinant of people’s life chances in the United States. And there are even Marxist sociologists — who you’d expect would have the most issue with this proxy — who’ve come around to the importance of the college divide.
In the first analysis, we were looking at racial and class inequality in prison admission. Here, we only have measures of education; we don’t have measures of income. But in the second two analyses — of people’s likelihood of having a family member imprisoned and people’s likelihood of living in a high imprisonment neighborhood — we had both education and income. And the results were almost identical. And so in this particular case we’re not especially concerned about using education as a proxy for a class, even though we acknowledge that the two concepts are different.
Q: One of the problems you have in doing this research is not only trying to figure out what serves as a useful proxy, but how to extract the information from whatever data you’re getting from the prisons or other systems. How did you manage this giant data sample that you had?
Roehrkasse: There are three key quantities that we’re trying to measure in this study, and we use three different datasets to measure each of those. Each of those datasets has its own unique value, and some serious limitations.
The first quantity we’re interested in is the likelihood that people enter prison. You might think that’s a really straightforward thing to measure. But it turns out that there’s actually no national data that are publicly available that disaggregate rates of entrance into prison by people’s race and ethnicity or their educational attainment. And so for people who are interested in these kinds of inequalities, a really useful and common resource is what’s called the National Corrections Reporting Program. Unfortunately, this resource is restricted in access, because it involves individual-level records of imprisoned people, so the data are pretty sensitive. But for those people who are interested in these kinds of questions, this is really the most important resource available. These are administrative data, and, unfortunately, they represent the voluntary contributions of different state prison systems to this overall program. In any given year, the NCRP doesn’t actually include all state prison admissions. So an important assumption of our study is that the contributing states in the years we examine are more or less representative of the country more broadly. It’s also important to say that the NCRP no longer includes federal prison admissions. Federal prisons make up a small proportion of the total prison population in the United States, but it is by no means a trivial proportion.
A second quantity that we’re trying to understand is the likelihood that someone has had a family member go to prison. And people can use any number of different resources to do this. People have used the Fragile Families study before or the Panel Study of Income Dynamics. We use a new survey that’s designed specifically to measure this quantity. It’s called the Family History of Incarceration Survey, or FamHIS.
The third quantity we’re interested in measuring is the likelihood that people live in a neighborhood with a high imprisonment rate. This is really challenging, because people aren’t usually imprisoned in the neighborhoods where they were living before they went to prison, and geo-locating prisoners back to the neighborhoods where they came from with any detail can actually be quite difficult. To do this, we use a resource that’s actually pretty underutilized, called the Justice Atlas of Sentencing and Corrections. This is another administrative dataset that compiles information from about 20 states, and it allows us to geolocate people in state prisons back to the specific census tract where they resided before they were imprisoned. We use census tracts, which on average have about 4000 residents, as a proxy for neighborhoods. And we use these data to calculate imprisonment rates for census tracts in these 20 states. Then we use census data to put people of different races and ethnicities and educational groups into neighborhoods to understand their likelihood of living in a high-imprisonment neighborhood. Then for all three of these experiences—prison admissions, family member incarceration, and neighborhood incarceration—we calculate the rates at which people of different ethnoracial groups and educational groups have these experiences. And then to measure inequality, we look at the ratio of these different rates across different groups.
Q: Another aspect of the complicated nature of this research is the temporality problem you have. When you’re looking at prison admissions, these are people who are entering the system. This is not representative of the body of people who are currently imprisoned as a whole. But then you’re asking people about the experience over their lifetimes, whether they’ve known someone who is incarcerated. How do you disentangle these different temporal aspects in this research?
Roehrkasse: This is a really important point. Our study is focused on prison admissions, specifically the rate at which people in the population enter prison in any given year. And this is a pretty different quantity from the proportion of the population that’s imprisoned at any given point in time. Generally speaking, prison admissions are much more volatile than prison populations, because they’re going to be more responsive to economic, social, and political changes. For example, a policy that diverts people away from the criminal justice system would have a pretty immediate impact on prison admission rates, but only delayed effects on the prison population, because that population reflects not only that recent policy, but the cumulative history of decades of previous policies, rates of imprisonment, sentencing, corrections, etc. What that means is that if we were to redo our study examining prison populations, instead of prison admission rates, some of the changes in inequality that we document would probably be a bit more muted. But what that also means is that if the trends we document in our study continue, we should expect to see similar changes in the prison population over time. There other aspects of our data—like the fact that the FamHIS survey captures whether a person’s family member has ever been imprisoned—that incorporate this whole cumulative history of incarceration over the last several decades, that we’re just limited in our ability to deal with.
Q: That points us back to one of the key topics people talk about with mass incarceration, which is the War on Drugs. How did the War on Drugs become so central to the conversation around mass incarceration, and how did your research complicate this story?
Muller: The paper itself is not directly about the War on Drugs, but the War on Drugs has become a key part of debate over mass incarceration. On the one hand, if you look at a point in time, the number of people who are in prison strictly for drug offenses is actually quite small. People often are critical of the argument that the War on Drugs was a key part of mass incarceration, given the small proportion of people who are in prison for drug offenses.
On the other hand, if you have people going into prison for relatively short sentences, that is going to mean that for people’s experience of having ever gone to prison, the relative importance of the War on Drugs is likely to be quite a bit larger. So, the temporal aspects we’re talking about have a particular relationship to the War on Drugs.
Alex pointed to these extreme disparities in incarceration during the mid-1990s, even within educational groups. I haven’t seen a study that’s nailed this down, but I think it’s unlikely that some part of that spike does not have anything to do with the War on Drugs. Some of the spike in the racial disparity in the prison admission rate in the 90s almost certainly was related to the War on Drugs. And so the War on Drugs is quite clearly is an important part of the story. How important it is really depends on which aspects of mass incarceration you’re trying to look at — whether you’re looking at the number of people in prison and the proportion of them who are in for drug offenses, whether you’re looking at people who’ve cycled through prison, and how many of them have been imprisoned for drug offenses, and whether you’re looking at racial disparity. I think you’re going to get a slightly different story, depending on which of those quantities you’re focused on.
Q: You’ve also done research on how factors like the labor market play a central role in how we explain rises in imprisonment and mass incarceration. Can you tell us more about this relationship?
Muller: First let me step back and talk about the previous state of the literature on the causes of mass incarceration, then I’ll talk about my own research. To be honest, I’ve been working on this topic for a while, and the longer I’ve worked on it, the more complex the answers have gotten about what the sources of mass incarceration are.
The broad contours are set out in a book by a sociologist named Bruce Western called Punishment and Inequality in America, which came out in 2006. Those main causes are still pretty widely accepted, even though there’s been a lot of important work to appear since that book was published. Western focuses mainly on economic and political causes, things like the collapse of urban labor markets, the related rise in crime, the urban uprisings of the 1960s, and then the politicization of crime that increased the chance that all of these changes would receive a punitive response. In the following years, we saw sentences increase, and we saw a greater willingness among prosecutors to pursue incarceration in cases where they might not have in the past. That’s an oversimplified summary, but it captures the main currents, and though people will disagree about the relative weight to place on any one of those causes, very few would say they’re wholly unimportant.
To give broader context, one of the main motivations for my work on incarceration — and for my work in other areas — has been the idea that, in my view, too often in sociology we begin our studies of racial inequality in the 1960s, and that leaves out a lot of really important historical context. We forget, for example, that for much of US history, Black Americans worked primarily in agriculture, not just during slavery, but for almost a century after the Civil War. Once you recognize this fact, a lot of otherwise puzzling features about long-run patterns in the Black incarceration rate begin to make more sense.
To take one example, there’s a popular argument that after the Civil War, incarceration became a kind of functional replacement for slavery. This is different from the argument that the form that incarceration took closely resembled slavery, which is an argument that has a lot of support, especially if you’re looking at the convict lease system, chain gangs, or things like that. But if you’re looking at the functional replacement argument, it’s hard to square with the fact that the Black incarceration rate in the years after Reconstruction was actually lowest in the counties that had depended most on enslaved labor before the Civil War. A lot of people are surprised when they hear this fact. But it becomes less surprising once you recognize that slavery and sharecropping were systems of economic exploitation, in addition to systems of racial domination. Both slaveholders before the Civil War and planters after the Civil War depended heavily on Black Americans’ labor. What that means is that, unless they could use the labor of people in prison, they had strong reasons to try to keep workers out of prison rather than in it. One of the key underappreciated ways that they did this is that planters often would go to courthouses, and they would offer to pay the fines of any people who had been convicted. The person then had to pay off the “debt” by working on their land. This system of peonage allowed planters to reestablish a coerced labor force after the Civil War. But it also had the side effect of lowering the Black incarceration rate in the Cotton Belt. So rather than see a relatively low Black incarceration rate in the Cotton Belt in those counties where slavery had been most prevalent after Reconstruction as a sign of the region’s mercy, we should instead see it as a sign of Black Americans’ continuing unfreedom outside of the prison in the years after the Civil War.
There’s an additional puzzle that this way of looking at things helps to solve. Often, critics of the functional replacement argument — critics of the idea that incarceration was a replacement for slavery — will say, “Well, if slavery and mass incarceration are connected, why does mass incarceration take off a century after slavery ends?” For me, a key part of the answer to that question is that cotton harvesting was almost fully mechanized between 1950 and 1970 — the two decades that precede the start of the prison boom. A lot of work has focused on the effects of deindustrialization, but there’s been much less of an emphasis on the collapse of agricultural employment. This is particularly important because the effects of the collapse in agricultural employment on Black men’s labor force participation were much larger than the effects of deindustrialization.
Q: That’s fascinating because it points us to this question of the relationship between these different labor markets and ties it into other historical phenomena that we might be familiar with, like the Great Migrations. As we switch towards the 1970s, how was the labor market shift related to the rise of mass incarceration?
Muller: There are three main ways we could think about this. Here I’m more synthesizing previous work, rather than drawing on my own, but we had a massive collapse in the share of young Black men who were working in agriculture. In 1940, about a third of young Black men worked in agriculture. By 1970, it was lower than three percent. It was a dramatic shift. I don’t know of any research looking directly at the effects of this mechanization of cotton harvesting on both changes in crime and changes in imprisonment, but there’s a lot of work looking at other shocks to the labor market and showing quite clearly that those are related both to rates of crime and to rates of imprisonment. That’s actually something I’m working on right now.
Secondly, one of the main responses to the mechanization of cotton harvesting was the second Great Migration. There was a huge political backlash to this migration. Ellora Derenoncourt, an economist who was at Berkeley until very recently, has shown how the second Great Migration led to increases in police spending, in homicide rates, in the Black incarceration rate, and in reductions in spending and other types of public goods. Ellora’s work shows clearly how this second Great Migration was related to the onset of mass incarceration.
Thirdly, there have been economic historians who have argued that the mechanization of cotton harvesting and the second Great Migration created a material foundation for the rise and the emergence of the Civil Rights Movement. Of course, a lot of the literature on mass incarceration discusses how there was a political backlash to this movement and focuses on this as a key component of the politicization of crime — one of the key ingredients in the rise of mass incarceration.
So, it’s through a bunch of different paths, but I do think many of these causes that other scholars have focused on are related to this massive decline in agricultural employment that happened mid-century in the United States.
Q: What can scholars and policymakers learn from your research on the complicated relationship between race and class?
Roehrkasse: Part of our analysis is aimed at decomposing racial and class inequality: overall, racial inequality in mass incarceration appears in part to reflect some underlying disparities in educational attainment. That’s an important fact to understand.
But one of the main goals of our study, and I think one of its main successes, is to show that racial and class inequality cannot be disentangled. And that’s because they’re mutually constitutive. That can sound kind of hand-wavy, but we make our best effort to measure this as concretely as we can. We show that, irrespective of one’s education or income, Black people are much more likely to have family members or neighbors imprisoned. This can seem somewhat at odds with the fact that we’re simultaneously documenting that there’s been this shift toward much greater educational inequality in prison admissions.
We think, though, that a really important factor that can reconcile these two seemingly contradictory facts is that, as a result of racial segregation and racial discrimination, an important feature of being Black in America today is that, irrespective of your class position, you’re much more closely connected to poor people. What that means is that the scale of racial inequality really can’t be fully appreciated without reference to the ways that social networks and social environments translate these growing class disparities into racial disparities.
Rather than being competing forms of inequality, race and class are really intersecting dimensions of domination. And for researchers, for activists, and for policymakers, the more we can do to understand that, the more successful we’ll be in our efforts to combat mass incarceration.