The First Step Act: Is It Working and What’s Next?

Event Video

Listen & Download

The First Step Act of 2018, passed as the result of bi-partisan efforts during the Trump administration, aimed to reduce the population of those in federal prison and to limit some federal prison sentences. Over the years some have contended the act is working well, while others argue it has only partially delivered on its goals or it was flawed from the start. Now, as the act recently celebrated its 5-year anniversary, join us for a panel discussing the First Step Act, its impact, legacy, and future.


  • Stephanie Kennedy, Policy Director, Council on Criminal Justice
  • Rafael A. Mangual, Fellow and Deputy Director of Legal Policy Contributing Editor, City Journal, The Manhattan Institute
  • (Moderator) Vikrant P. Reddy, Senior Fellow, Stand Together Trust


As always, the Federalist Society takes no position on particular legal or public policy issues; all expressions of opinion are those of the speaker.

Event Transcript



Chayila Kleist:  Hello, and welcome to this FedSoc Forum webinar call. Today, February 27, 2024, we’re delighted to host a discussion on “The First Step Act: Is it Working, and What’s Next?”


My name is Chayila Kleist, and I’m an Assistant Director of Practice Groups here at The Federalist Society. As always, please note that all expressions of opinion are those of the experts on today’s call, as The Federalist Society takes no position on particular legal or public policy issues. Now, in the interest of time, I’ll keep my introduction of our guests today brief, but if you’d like to know more about any of our speakers, you can access their impressive full bios at


Today, we are fortunate to have with us Dr. Stephanie Kennedy, who is a Policy Director at the Council on Criminal Justice. Dr. Kennedy has spent more than a decade conducting research concerning justice-involved people and families, with a particular focus on women and mothers. Her scholarship focuses on community supports to reduce criminal behavior and victimization, generating data-driven criminal justice system and policy practice reforms, and other such research issues.


Prior to joining CCJ, she was Director of Research Dissemination for an academic research institute focused on developing and testing well-being-based behavioral health interventions for currently or formerly incarcerated individuals, correctional staff, and law enforcement officers. Dr. Kennedy holds a Master of Social Work and a Ph.D. in Social Work Research.


Also joining us today is Rafael Mangual, who is a Fellow and Deputy Director of Legal Policy at the Manhattan Institute, as well as a contributing editor to the City Journal. He has authored and co-authored a number of Manhattan Institute reports and op-eds on issues ranging from urban crime and jail violence to broader matters of criminal and civil justice reform.


His work has been featured and mentioned in a wide array of publications, including the Wall Street Journal, The Atlantic, New York Post, Boston Herald, and City Journal. Mr. Mangual has also made a number of national and local television and radio appearances on outlets such as Fox News, C-SPAN, and Bloomberg Radio. Prior to joining the Manhattan Institute, Mr. Mangual worked in corporate communications for the International Trademark Association.


Finally, joining us as our moderator for today’s discussion is Vikrant Reddy, who serves as a Senior Fellow at Stand Together Trust, specializing in the area of criminal justice reform. Mr. Reddy previously served as a senior policy analyst at the Texas Public Policy Foundation, where he managed the launch of TPPF’s national Right on Crime initiative in 2010.


He’s worked as a research assistant at the Cato Institute, a judicial clerk to the Honorable Gina Benavides in Texas, and as an attorney in private practice. He’s a member of the State Bar of Texas and also an appointee to the U.S. Commission on Civil Rights Texas State Advisory Committee. Mr. Reddy’s research and scholarly opinions have appeared in a range of national media outlets, including USA Today, National Review, The Federalist, and others. And I will leave it there.


One last note. Throughout the event, if you have any questions, please submit them via the question and answer feature that’s found at the bottom of your Zoom screen so they’ll be accessible when we get to that portion of today’s webinar. With that, thank you all for joining us today. Mr. Reddy, the floor is yours.


Vikrant P. Reddy:  Terrific. There are so many things to be discussed and debated regarding the First Step Act, what should have been in it, whether it should have been passed, whether it has worked, whether there should be a Second Step Act. All those things will be discussed today.


But it occurs to me that one thing that cannot be debated is that this was the most significant—at the federal level—the most significant criminal justice reform passed in a generation. It was a very big deal; it was a very big story.


Now, having said that, it also occurred to me. I just kind of tried to do the math in my head this morning. Assume that you are a first-year law student, and you are on this webinar right now. When the First Step Act was passed, you were still in high school, and maybe you don’t really know what it was because you weren’t following politics or public policy at the time.


I think before we really dive into some of the deeper questions about what it has achieved and what the future is, we should just speak very directly about what it was, what is in it, what it was intended to do. And I think probably the best way to do that is to give the floor to you, Stephanie, because you’ve got a presentation that you give that covers a lot of these basics and gives us that overview.


Stephanie Kennedy:  Fantastic. Thank you so much. Bear with me for one second while I share my screen, which is, of course, being difficult. Okay. You can see my screen, correct?


Vikrant P. Reddy:  I can.


Stephanie Kennedy:  Fantastic. So, as stated, I’m Stephanie Kennedy. I’m the Policy Director for CCJ. And I want to talk a little bit about an analysis we did that really explored the -- what’s going on with the First Step Act and how it’s affecting folks in our federal prison system.


So, just for those of you who are new to this, the First Step Act was passed in 2018 with strong bipartisan support, and implementation began in 2019. It’s important to remember that the FSA affects only those incarcerated in federal prisons—roughly 160,000 people overall. And the goals of the Act are to identify people who can safely be released, reduce recidivism through enhanced program participation and skill training, improve prison conditions, and address a range of sentencing issues. The FSA mostly targets those at minimum or low risk for recidivism, although there are provisions which apply more broadly.


There are a lot of provisions in the FSA, and they fall into three main buckets. So we have correctional reforms, which include using a risk-and-needs assessment, sentencing reforms, which include modifying some mandatory minimum sentences for some, release mechanisms, including increasing good time sentence credits and the way that they’re allocated. So in our analysis, we looked at the 29,946 folks who are being tracked for recidivism as reported in the 2023 annual FSA report published by the Office of the Attorney General.


As you can see, these folks were released under a range of release codes reflective of the multiple provisions of the First Step Act. Our colleague, Dr. Avinash Bhati, of Maxarth Consulting used data drawn from the sample to construct three separate analyses to explore how the First Step Act impacted both time served and recidivism. We received some data from BOP to support these analyses, but in general, these are drawn from values published in the official annual reports. You can read the full analysis and methodology sections that pair with each on our website, and I’ll drop that link in the chat later.


And so the first finding is about the impact of the FSA on time served. So how much time folks are actually spending in prison, specifically on the share of the imposed prison term that people are serving under the FSA.


For this analysis, we used the roughly 20,000 folks who were released under the FSA in 2022. It’s about two-thirds of all releasees. And then we constructed a comparison group of people released from BOP with the same charge of conviction before the FSA was passed.


So prior to the FSA, folks were serving just under 90 percent of their imposed prison term. The 2022 FSA releasees served roughly 82 percent. So that represents a drop but a modest one of about seven percent. To contextualize, on average, people released under the FSA advanced their release date by about five months. So you can really see the modest impacts here. Release dates were advanced by less than 1 year for 92 percent of all of those released under the First Step Act in 2022.


We used the full sample -- the full FSA release sample to examine recidivism. So, as I showed you in that first pie chart, there are 29,946 people who were released under the FSA from 2020 to 2022 or to January 2023 and who are being tracked for recidivism. You might have heard the recidivism comparison in the news where the 12.4 percent recidivism rate for FSA folks was contrasted to a 46.2 percent general recidivism rate. But it’s not an apples-to-apples comparison, as eligibility for release under the FSA is fairly targeted.


So we constructed a comparison group of similarly situated people who were released from BOP prior to the implementation of the First Step Act. We matched on risk assessment classification, as the FSA targets folks at minimum and low risk, and factored in time since release, as roughly two-thirds of all of those FSA releases happened in 2022.


So while the overall BOP recidivism rate is 46.2 percent, we estimated a rate of 19.8 percent for our similarly situated folks who were released prior to the FSA. So the 12.4 percent recidivism rate reported for people released under the FSA is roughly 37 percent lower than our similarly situated comparison group.


We then used these two similar groups to estimate the number of arrests that each could have been responsible for over three years. Just a quick reminder that recidivism, as defined by the Bureau of Prisons, is any arrest, regardless of outcome, or a return to federal prison. So people released under the FSA could have incurred about 4,300 arrests and their similarly situated counterparts could have incurred about 7,500 arrests, suggesting that FSA releasees incurred about 3,100 fewer arrests over 3 years.


We did receive a small amount of additional data from BOP, which allowed us to disentangle recidivism due to technical violations from that due to new crime. So 31 percent of the recidivism among folks released under the FSA was due to technical violations not related to new crimes. So this means that that 12.4 percent recidivism rate reported for FSA releasees actually drops to 8.5 percent when only looking at new criminal activity.


And so using these data, we were able to adjust the estimated maximum number of arrests due to new criminal activity incurred by folks released under the FSA from about 4,300 to just under 2,900. We did not have similar data on our similarly situated comparison group, so we were unable to conduct that analysis.


And so, just situating these arrest numbers in the context of the roughly 20 million national arrests that were made in 2020, ‘21 and ‘22, we found that arrests for new crime incurred by people released under the First Step Act would account for 0.014 percent of nationwide arrests—about 14 out of every 100,000. So the big takeaways here are that the First Step Act has resulted in modest drops to time served and does appear that folks released under the FSA have lower rates of recidivism.


I want to be very clear that this is not an evaluation of the FSA and that our findings are limited by their reliance on aggregated data. As evidence-based recidivism reduction programs become more widely implemented and available, the results that we present may change over time.


Vikrant P. Reddy:  Stephanie, before we jump into questions, Ralph, I want to give you the opportunity to just, at a broad level, offer some reflections and thoughts on the First Step Act. And you can also feel free to respond to anything that you saw in Stephanie’s presentation, of course.


Rafael A. Mangual:  Yeah, thanks. Look, I think the First Step Act is an interesting reform for us to reflect on and to continue reflecting on in the years to come. But I think the best way to characterize it really is kind of an exercise in the limits of safe decarceration. We’ve kind of heard a little bit of this from Stephanie’s presentation.


But as you can see, the individuals that are impacted by the First Step Act, it’s a very narrow group of individuals that pose a pretty low risk to society in general. And one of the things that has always troubled me about the rhetoric surrounding the First Step Act was the degree to which individuals talked about it as a first step, as a first step down a much longer road toward a broader decarceration project.


And I think to the extent that we’re going to see the First Step Act that way, then we have to answer certain questions about the degree to which whether it worked and the degree to which the current data that we have indicate that we can extend this more broadly without actually impacting public safety.


So, again, when I talk about this being an exercise in the limits of safety incarceration, it’s important to understand just how tiny the population of people incarcerated in the United States this actually impacts. We’re only talking about the federal population, which is only about 9 out of every 10 -- I’m sorry, 1 out of every 10 -- a little more than 1 out of every 10 prisoners in the country or in the federal system. And we’re only talking about a small share of that federal population.


I mean, if you look at the First Step releasees, you’re talking about close to 30,000 individuals. But when you dig into those numbers, you’re really only talking about 14,000 people who have been released who actually entered a Bureau of Prisons prison facility, right? A lot of the individuals who have been released under the First Step Act never actually entered prison. They were released straight from pretrial detention or pretrial supervision. So that brings the number down to about 14,000 individuals, which is less than 1 percent of the overall prison population in this country—about 1.8 individuals incarcerated in the U.S.


When you think about the First Step Act and what it was intended to do, I think the major components beyond extending and rendering retroactive some of the sentencing reforms, like the Fair Sentencing Act of 2010—the resentencing for crack cocaine offenses, etc.—the main thing that it was really supposed to do was to better identify the risk that individuals entering BOP facilities pose so that they can be housed more strategically and then be given a suite of programs that was a little more tailored to the risk that they posed so that we can reduce their likelihood of recidivism upon release.


But we don’t really know very much about the degree to which those particular programs are doing what they’re intended to do. We haven’t had really the time to do the kinds of rigorous analyses that we would need to do to evaluate the integrity of those programs, to answer the question of whether exposure to certain programs is reducing recidivism likelihood to some significant degree. And there are a lot of other limits on our ability to say with certainty at this point in time whether the Act is working.


One of the things that’s important to understand about the First Step Act is that it was very slow to implement a lot of these changes. So one of the things -- it took a long time for the risk assessment to get developed and off the ground. It took a long time for a lot of these programs to get off the ground and implemented and for inmates to get into them and actually go through them. And we really only have about a year’s worth of time between now and when release dates starting getting recalculated based on the earned time credits for program participation, which I believe happened in January of 2023.


And so all of that has to be taken into consideration when you view some of the measures that are intended to indicate success. And one of those measures, of course, is recidivism, which we’ve heard about. And I think Stephanie covered this really well. But I think, again, the most important thing to understand is that the broad recidivism measures that are getting touted -- for example, in Newsweek, there was a piece written by one of the first beneficiaries of the First Step Act, Matthew Charles, and I believe a member of Congress, but I can’t remember who it was.


But they were, again, touting this 12.4 recidivism rate compared to the 43 or 46 percent recidivism rate for the Bureau of Prisons generally. But again, that’s not an apples-to-apples comparison. This is a much narrower group in terms of First Step Act beneficiaries than the general prison population.


The other thing that we have to keep in mind, too, is that the observation period here for recidivism is relatively short, right? As you saw, the biggest cohort of individuals released in under the First Step Act was released in 2022. Generally, you see a lot of recidivism measures that are based on a three-year observation period. But what we know from some of the state prison release cohort data that the Bureau of Justice Statistics has released in recent years, when you extend that observation period to 10 years, you find a pretty large increase in the number of rearrests that get generated as time goes on.


I can’t remember exactly what the rate is, but I think over a three-year period, it’s something like 67 percent of state prisoners who are released are rearrested for a new offense. When you extend that to 10 years, it goes up to about 80 percent to 83 percent, right? Extending that time horizon is going to be important for us to know whether whatever changes can be attributed to the First Step Act in recidivism are sustainable over time.


The other thing that we know from the release data -- the recidivism data, is that the higher -- the few higher risk individuals that do -- that have benefited from the First Step Act do have higher recidivism rates than the lower risk individuals. But we don’t see huge differences in recidivism data between and among individuals who have actually participated in the evidence-based recidivism reduction programs. So that’s, I think, going to be one of the biggest things that we want to look at moving forward is to see whether or not these programs are actually doing their job in a way that’s cost effective.


Another thing, too, is when you’re looking at some of the release data, I think it’s important to just take into consideration that individuals released under the First Step Act are released into a somewhat different environment than individuals who have been released in years past. The likelihood of arrest in the United States has gone down significantly as the number of police on the beat has decreased, as clearance rates have gone through the floor in many cities. You’re just less likely to be arrested conditional on a crime having been committed sort of post-2020 than you were in years prior. And so I don’t know how you take that into account, but it’s something to think about.


And one thing I will say just in response to something that Stephanie brought up on the technical violations front, it -- I think it’s worth breaking those individuals out. But it’s also worth remembering that some share of people who were returned to prison for a technical violation still did have a new criminal charge. And the technical violation was pursued in lieu of pursuing the new criminal charge.


And also, a good number of technical violators—at least in the federal system—these are individuals who have racked up multiple technical violations before they’re actually revoked, right? So it’s important to understand this isn’t just one failed drug test or one missed meeting. Usually these individuals are considered to be absconders. They have multiple factors in terms of technical violations that they’ve racked up.


So there was one study that was actually released by the U.S. Court Administration looking at federal probationers who were drug offenders. And it was about a third of the cases of individuals who had been returned to prison for a technical violation had at least five technical violation factors present and at least four combinations of factors. So that could be, again, missing a meeting, failing a drug test, not participating in a program, etc. So it’s just worth noting that as well.


Like I said, look. I do think that we have to recognize that not just in the federal system but in the state system, there is some subset of the population whose incarceration either no longer or never did serve a legitimate penological end, i.e., that there are individuals behind bars who don’t belong there, who don’t really need to be there in order to keep society safe. And we should identify who those individuals are and try to secure their releases in a responsible way.


But we have to understand that that project is a much smaller scale project than I think a lot of people in the criminal justice reform movement want to pursue. We’ve heard a lot about things like the CUT50 initiative, for example, which wants to reduce the incarceration rate in the country by half. You can’t do that just by targeting the kind of low-level risk propositions that the First Step Act is currently targeting. And it’s important for us to just really internalize that.


Vikrant P. Reddy:  I have a million things I want to ask. I’m going to start with one idea that both of you in different ways presented and seemed to agree on, and that is that, yes, we’re at the five-year mark. But in some ways, it is still a little too soon to really fully evaluate the First Step Act.


And my question then, I guess, is when is the right time to evaluate it? Ralph presented this interesting point about how as the years go by, the recidivism rates of released individuals take higher and higher. But I don’t know.


As I think about that, that sort of seems obvious and doesn’t -- I’m not sure that says what we think it says. So, for example, just hypothetically, assume a 20-year-old who gets locked up behind bars is there for a year. He comes out, and he’s 21. The day he gets out of prison, he gets in a bar fight. We would all probably say, “I’m not sure that prison really helped that guy get better. He’s not rehabilitated.”


However, suppose this guy gets out at age 21, and he doesn’t get in a bar fight again until he’s 50—odd to think of a 50-year-old guy get in a bar fight, but assume it happened. Thirty years went by, and this guy kept his nose clean. I don’t think any serious person would say, “What do you expect? He’s the next con.” I mean, that just doesn’t seem to make sense.


So, yeah, it’s obvious that every year that goes by, the likelihood of somebody getting arrested and getting involved in the criminal justice system again gets higher. There has to be some reasonable number that we can all more or less agree on when we think, “Okay, this is a good time to take the recidivism rate seriously.” Stephanie, I know there’s a technical answer to this, which maybe you could speak to, but then also offer some reflections on kind of the theory and philosophy of the statistics here.


Stephanie Kennedy:  I will do my best on that. I think a common kind of long-term recidivism window is usually between three and five years. And a lot of the desistance literature suggests that by year seven or eight, that rates of contact with the criminal justice system for folks who have previously incarcerated tend to come down to be about parity with the never incarcerated general population.


I think to your point, I think we are in an early phase of evaluating the First Step Act because it was implemented right as COVID came down. And so a lot of the provisions were not able to be implemented, specifically with this evidence-based programming.


And so I think in our first analysis—I pulled it up—we found that roughly 54 percent of the individuals released under the FSA completed at least one evidence-based recidivism reduction program. And so that feels like a very low number. And I understand that with the complications to making such a big change to a huge system that spread out across the country as well as having COVID really confine the number of people that could come in as volunteers and hiring and kind of all hands on deck as far as trying to ensure people’s health and safety, that program implementation just has not been as fast as we wanted it to be.


But I think, as in 2022, the releases really picked up, and programs began to be implemented. So I’m eager to see the next report that should come out roughly in April that talks about the gains made in 2023: How many folks were released? What kind of programs did they get? How were these other provisions of the First Step Act applied to them because one of the major provisions—well, not major—one of the provisions of the First Step Act is to try to move folks to a facility closer to home so that they’re able to be nearer to their community, can have a visit from their family, and kind of reconnect those social bonds?


And with COVID that just stopped, which makes sense. They didn’t want to move folks around if they didn’t need to to reduce risk. But I think as we come out of the COVID era, it’ll be really interesting to see how that shakes out. I will say our analyses are based on aggregate data. In order to conduct any kind of evaluation, you need individual-level data. And so that is a limitation right now for publicly available sources.


Vikrant P. Reddy:  Ralph, what do you think? At what point would you start to have some confidence in the recidivism numbers we’re looking at?


Rafael A. Mangual:  Yeah. I mean, I think when the majority of people who are getting released under the First Step Act have actually been given programming that was tailored based on their risk assessment and they’ve been out for at least five years, I think that’s a point at which we can have something really useful to say that we can base future policy decisions on.


Again, I think it’s important that the biggest chunk of individuals who have been released under the FSA were released in 2022. It’s about a year and a half ago—even less than that. So they just haven’t been out very long. And again, if you look at the most recent annual report that BOP put out on the First Step Act, you’ve got 14,000 -- a little over 14,000 inmates who did not complete one single evidence-based recidivism reduction program, which is about half of the 29,900 number of individuals that were released on the FSA.


And again, a big reason for that is that a huge chunk of the individuals who didn’t complete any programs never actually entered a Bureau of Prison, either because they were just sentenced to time served in pretrial detention, or they just weren’t in long enough.


I just think it’s important to have a data set where you’re looking at individuals, A, who’ve actually done at least a year in the BOP, in an actual prison, and who have been given a risk assessment and given programming based on that risk assessment, have completed that programming, and then have actually been released for some significant period of time. I would like to see five years, but at least three years, I think, would start to give us some interesting indications.


And then the other thing, too, is it’s going to be important, I think, for us to disaggregate based on the type of programming that the individuals have been exposed to and also disaggregate based on the charges that individuals are facing, right? I mean, again, huge. I think it’s like—I don’t know—75 percent of the individuals released were either a low risk or a minimum risk in terms of their pattern assessment. So, again, there’s a limit here, right? This is what I meant by saying this is an exercise in the limits of safe decarceration, right?


How effective are these recidivism programs with anyone beyond that risk level? Yeah. That’s the long-winded answer to the question. Once we’ve got five years of release data from individuals who actually served time in prison, we’re given these assessments and programming based on the assessments, and then that’ll give us the opportunity to start breaking out these programs from one another, figuring out which programs work, which programs need to be rejiggered.


How do these programs work with different categories of offenders, either based on the conviction charge or based on their initial risk assessment at intake? And how quickly do those risk assessments change over time, right? I mean, that’s interesting, too.


These evidence-based recidivism reduction programs are also intended to improve behavior inside the system. To someone who enters at a high-risk rating, are they leaving at a low-risk rating or a minimum-risk rating? And to what degree can we attribute that to the programming that they’ve been exposed to? I think these are all going to be questions that need to be answered definitively before we can really say whether the FSA has been a success or not and certainly before we can start making decisions about future policy initiatives based on the FSA data.


Vikrant P. Reddy:  Before we jump into questions from the audience—and there are a lot—let me ask a question that I bet some people are thinking. The First Step Act gets passed in 2018. First releases begin to happen in 2019. Then 2020, it’s simply undeniable that violent crime rates in the United States begin to tick upward. Can any connection be drawn, or is that way too simplistic and facile?


Rafael A. Mangual:  I don’t think so. I mean, again, just because of how small the number of individuals you’re talking about, I mean, yeah. We saw a massive violent crime rate. But I would think that the state prison population decline over that period probably has a bigger impact in the state jail population.


We saw a 25 percent decline in the jail -- or a 15 percent decline in the jail population in 2020. We saw an acceleration in the decline in the state prison population in 2020 and ‘21. I think those things probably had a lot more to do with the overall crime increase than 14,000 individuals spread throughout the country.


Vikrant P. Reddy:  You were nodding your head, Stephanie. You agree?


Stephanie Kennedy:  Yeah. And I think most of the people who are in federal prisons are not in for violent crime. And so the First Step Act, certainly, I don’t think these releasees -- there were very few of them that happened in 2020. And so these releases had nothing to do with what was happening in communities, for sure.


Vikrant P. Reddy:  Okay, just jumping into questions here. There was one that caught my attention about programming in the BOP. And we have a former DOJ official who’s saying that having worked within DOJ, it was actually very difficult to find programs that had a proven effect on recidivism. And I think we should just take that question very broadly. What do we know about the kinds of programs that help reduce recidivism? And to what extent does the First Step Act -- has the First Step Act enabled DOJ and BOP to adopt more of those kinds of programs, if any?


Rafael A. Mangual:  Yeah. I mean, look. I think it’s hard to say. I think that a fair reading of the literature on recidivism reduction at least leads me to the conclusion -- or a conclusion similar to that drawn by Robert Martinson in the 1970s, which is very, very little work.


We don’t really know how to reliably reduce recidivism—certainly not at scale. Do we see some programs that show promise? Yes. Do those programs often prove difficult to duplicate and to scale up? Yes.


One of the things that you often see in a lot of these programs that have impressive numbers at the outset at least is that you’re often dealing with some selection bias issues, right? So some of these programs exclude certain categories of offenders. Some of these programs require offenders to opt in, affirmatively applies, which means you’re dealing with a subset of offenders that have already taken that initiative step and shown that that level of interest that might indicate that they’re a lower risk proposition to begin with.


But they’re also very intensive. A lot of the programs that tend to succeed—at least in the short term—are programs that involve a lot of one-on-one therapy, a lot of small group therapy. That’s difficult to scale across a prison population as large as the prison population in the United States.


The other thing that you see, too, is you’ll see success over a short observation period: 12 to 24 months. Extend the observation period to 48 months, and you start to see a regression back toward the mean, which means that whatever gains are attributable to the program are often hard to retain the further and further removed individuals become from the program. And, of course, you run back into the scaling problem. If you’re talking about a program that ultimately needs to get administered over a very long period of time, it’s just not feasible.


And also, we don’t know the degree to which these things could work as alternatives to incarceration because we don’t know what the compliance rate would be like outside of a restrictive setting. It’s one thing if someone doesn’t really have much to do and is going to group therapy and showing up every day. It’s another thing when that person is confronted with the choice between group therapy and, say, hanging out with their friends or going to a party or whatever it is.


So yeah. I mean, again, I think there’s a huge limit. I don’t think the FSA EBRRs have been in effect long enough for us to really have any useful data to say one way or the other whether they’re initially showing promise. And I think that’s just a function of the fact -- and that’s going to be a problem, I think, moving forward. But that’s just a function of the fact that recidivism reduction in general is a very difficult thing. And I think the data that we have about other programs over the years and decades lends itself to that conclusion, which is that it’s just a really hard thing to do.




Stephanie Kennedy:  I think it’s important to remember that I think we locate recidivism in the behaviors of a person who has left prison. But the recidivism construct, really, it’s about that person’s interaction with their community, with law enforcement, with their family, with work, with their supervising probation or parole officer.


And so there’s a lot of factors that go into recidivism that kind of talk about this sticky wicket of, “We’re looking for -- well, if we give you program X and we teach you how to stop thinking in this criminal way, then that will immediately translate into behavior Y that we are wanting.”


But oftentimes when you’re talking about personal change, behavioral change, lifestyle change, those things take a lot more time. And that person, if they’re in a community that has a higher police presence, now, they have a criminal history, people know who they are, they’re just -- they’re more likely to come into contact with law enforcement, which is a violation of your supervision, and to kind of face a host of other challenges. But I totally agree with you.


It is very difficult to implement programs that require fidelity to a model. That’s the kind of evidence-based model is you have a program. You test it. You do long-term outcomes, and then you have to replicate that in the way that the creators of that program intended over and over in this huge lumbering system with all these folks. And you’re dependent upon volunteers and correctional staff in order to lead these programs instead of the developer or someone who really knows how it works.


And yeah. I think your point, Ralph, about could we use some of these programs as potential mechanisms for pretrial -- for pretrial diversion as an alternative to incarceration is a really interesting one. And I would love to test that hypothesis and see if that could be a safe way to get folks what they need in a community setting while still giving them the structured support to get to that program, to get to work, to manage family responsibilities, and to manage the demands of social life that we all have to manage.


Vikrant P. Reddy:  I thought it was really interesting, Ralph, how in your answer, you talked about the evidence we do have that what tends to work is a lot of one-on-one counseling, which is obviously very time-consuming, very expensive. Therapists don’t grow on trees.


But I just want the audience to know I happen to know that the Manhattan Institute has actually written a great deal about this and about the ways in which maybe some more resources need to be poured into the -- into our—how should I phrase it?—well, just into our prison system. But it’s not a question of necessarily building more prison space, but it’s about getting the right therapists and the right programs and the right folks in there.


And I think that nuance is sometimes lacking in our public conversation about prison incarceration policy, which is either put more money, make it bigger, or take people out, make it smaller. It’s just more detailed than that.


Let me ask another question. This one’s a little detailed. I wonder if you guys have any thoughts on it. We have a question about the COVID home confinement program and whether or not there are any statistics about recidivism that come out of that and whether or not we can compare them to First Step Act statistics. Stephanie or Ralph, do either of you know anything about this? Have any thoughts?


Stephanie Kennedy:  A little bit. So the CARES Act folks -- and I’ve gotten this question before. It’s like, “How do the CARES Act folks figure into this recidivism analysis?” And my answer is they do, but it’s unclear based on the available data that we have exactly how they are counted.


So in that 29,946—that big recidivism sample—CARES Act folks are certainly numbered in there -- represented in there, but it’s unclear how many of them, as the CARES Act folks went from a BOP facility into home confinement. And so while they are on home confinement, they are not being tracked for recidivism in the same way because they are still under the supervision of BOP.


And so the only thing that I have seen is a published recidivism value of, I think, less than one percent. It’s less than a tenth of a percent. I forget. It was shockingly low. I think it was 3 or 4 people had gotten into trouble among the 3,500 or 4,000 who were being tracked. But I have not seen the actual data that back that -- from where those -- where that statistic was drawn.


Rafael A. Mangual:  Yeah. I mean, I think that’s right. Again, you’re talking about a very small sample. And then correct me if I’m wrong, Stephanie, but I do think that a lot of the CARES Act beneficiaries tended to skew older in terms of age because of their COVID risk. So that’s another thing to take into consideration, right? I mean, once you get above 50, the recidivism rate tends to drop off pretty significantly.


Stephanie Kennedy:  Yeah. I don’t have those data, so I can’t say that with confidence, but the CARES Act folks were certainly at minimum or low risk. And yes. In a COVID consideration, I would make an assumption—again, not backed by data—make an assumption that they were older and had been identified as minimum or low risk.


Vikrant P. Reddy:  There’s a question which, candidly, I don’t quite understand, but I have a broader sense of what would be a good discussion topic around this. I’m going to read the question. You guys can address it if you want, or you can wait for my reframing. The question is, “The political realm is ‘doing its part on early release with problematic no-bond release as one example.’ Is this discussion area one driver of the First Step Act’s original momentum?”


Rafael A. Mangual:  Yeah. I’m not sure I entirely understand the question either. But to the extent that it’s asking whether the First Step Act was sort of part and parcel of the broader decarceration push that helped get bail reforms over the line in a lot of states, I think it’s fair to say that the answer to that question is yes.


Does that mean that every supporter of the First Step Act is also a supporter of some of the more extreme sort of decarceration initiatives that we’ve seen—for example, in my home state of New York, where we are the only state in the Union that doesn’t allow judges to consider dangerousness and pretrial release? Yeah, I don’t think that would be fair to say either. But, yeah.


I do think that the First Step Act is kind of part of the broader reform push, which is why I think it’s so important to speak to the limits of the population that the First Step Act is really dealing with because, again, what we can learn from the First Step Act doesn’t really tell us very much about what we can do to safely decarcerate at the scale that a lot of people who would characterize themselves as sort of activists or sort of reform advocates that are pushing for decarceration on a broader level and want to see, which is kind of why I framed my initial comments the way that I did, right?


Again, I think it’s an exercise in the limits of relatively safe decarceration. And when you zoom out and understand that our prison population is as large as it is, I think that the data coming out of the First Step Act, I think, just reinforces the reality that there’s a very, very, very clear limit as to how much we can do on the decarceration front without impacting public safety.


Vikrant P. Reddy:  Stephanie, do you have any thoughts?


Stephanie Kennedy:  I think that hasn’t been tested. So I think when I hear you say it’s a limitation of decarceration, I think the First Step Act is -- it is a first step. And so I guess I have a different orientation towards it.


I’m like, “This is a great kind of way to test.” Can we shift folks who are at minimal or low risk, get them screened, identify their needs, put them in programs that we know are effective, give them the support that they need, move them closer to home, and shift them into home confinement sooner without impacting public safety?


And I think from this kind of preliminary aggregate analysis that I showed you, the answer is yes. I think that could be replicated in other systems, but it’s going to take some kind of thoughtfulness about how we could apply these principles to other systems and then build on them while testing. And so I think identifying smaller test models of are -- because there were folks who are at minimum or medium or high risk who have been released under the FSA. And so that can be done based on a warden’s discretion and some other kind of factors.


So there are folks who are, again, highly motivated, who have engaged in a change process and are demonstrating that their aptitude for success after release is very high, who are included in this. And so I think -- yeah. We need to find ways to start and see -- test other models for how we can safely move folks into alternatives to incarceration and get them the support services that they need so that they’re not engaging in either problematic behavior, like substance use, or engaging in crime.


Rafael A. Mangual:  Yeah. The one thing I’d just add to that is, yes, we definitely did have about 3,000 people who were either medium to high risk based on their pattern assessments who have been released under the First Step Act.


But I do think it’s important to note that for the medium-risk folks, you’re looking at a percent recidivating of 21 compared to 8 percent for low and 3.2 percent for minimum. And for the high-risk folks, it’s 34.4 percent, again, compared to 8 and 3.2, right? So I think that reinforces the limiting idea that I was talking about, given that their performance is already worse than the low- and minimum-risk folks.


And again, this would have to be an assumption, but my guess is that the higher and medium-risk individuals are also going to skew older and have served more time than the minimum- and low-risk folks who probably haven’t served that much time. So once you control for the differences in age, I would think that the disparity is probably going to grow. So just an important point to consider, I think.


Vikrant P. Reddy:  This conversation that the two of you are having about this last question makes me want to ask a little bit about our political context right now. You had a First Step Act. So in theory, there would be a Second Step Act or a Next Step Act.


Having said that, President Biden has had nearly four years to push on something like that. That hasn’t happened. President Trump, who may be President Trump again shortly here, has not necessarily been talking about doing anything like this on the campaign trail. I think there are people who are interested in doing more in criminal justice. I think they’re in the Senate—I would bet Senator Booker on the left, maybe Senator Lee on the right—that there are people who are interested in this.


But what do you think is the appetite for it at the moment? I can’t help but notice the -- just the absence of dialogue from Biden and Trump on the issue. Ralph, I think you follow this really closely and have probably got some thoughts.


Rafael A. Mangual:  Yeah. I mean, look. I think the appetite for it probably hasn’t changed that much within the community of advocates who have been pushing for this stuff over the years. What I do think has changed is the appetite for it among the general public.


I mean, if you look at the polling data, crime has, since the passage of the First Step Act, has really jumped in terms of priorities in terms of issues on the minds of voters. When you couple that with the crime increase that we’ve seen post-2020—which is, I think, probably best characterized as a continuation of a trend that started in 2015—we’re just seeing these kinds of initiatives decrease in terms of their popularity.


Again, that doesn’t necessarily mean that you’re not going to see any movement on the reform front. I think we continue to see plenty of movement at the state level, particularly in pretty deep, solid blue states that don’t have a lot of Republican opposition. Same thing for blue cities like New York and Chicago. I think they’ve continued to go down this road relatively unfettered by the more national-level shift in sentiment.


But when you’re talking about federal-level reforms, there’s a little more sort of parity in terms of power in Congress, and it’s a little harder to get some of these initiatives through. So for that reason, I don’t think we’re going to see any major criminal justice reforms come out of the federal system in the next five years. I could be wrong about that, but that would be my guess, whereas I think we’re going to continue to see lots of progress—to the extent that you want to call it that—on the state front.


Vikrant P. Reddy:  I hesitate to ask the question I’m about to ask because the answers could give us an entire other Teleforum, but I think we should talk about it. And that is that we’ve mentioned several times the violent crime increase in 2020 and in successive years. It’s also true that that increase has subsided dramatically.


I think, Stephanie—I’m maybe going to misquote a CCJ report here—but in something like the 30 largest cities in America, 28 have seen violent crime decreases in the past year. So it went up all of a sudden, and it’s coming back down, broadly speaking. And if it’s not the First Step Act that caused this horrible blip in violent crime, what did cause it? Surely, both of you got some thoughts. Stephanie, I’ll turn to you first.


Stephanie Kennedy:  That is, I think, a great topic of conversation and one that’s been hotly debated in the CCJ Crime Trends Working Group, really talking about what happened in communities and what the impact of COVID was on people’s routine activities and how that changed folks’ lives.


An increase in gun sales over 2020, I think. I do not have good data on this. The majority -- I can’t even say that -- many of which went to folks who already owned guns. But looking at that extra gun hypothesis, there’s also hypotheses that look at the impact of the murder of George Floyd and the protests that resulted in many communities across America.


And I think we don’t have a great -- there is no answer, as far as I understand it. There are a lot of hypotheses, and folks are doing data analysis and trying to test their theories and see if they can understand what happened, why it went up, and why it’s come back down.


Vikrant P. Reddy:  What do you think, Ralph? Are any of the hypotheses stronger than the others?


Rafael A. Mangual:  Yeah, I do think so. I mean, look. First, it’s important to note that even in the cities where the violent crime numbers are coming down, they’re still significantly elevated compared to 2019, right? So they haven’t gotten back to their sort of pre-pandemic levels.


I do think Stephanie raised an interesting point about the sort of shift in routine activities post-COVID. But what’s interesting about it to me, there was Maxim Massenkoff—I can’t remember how I pronounced his name—and Aaron Chalfin did a really interesting study that was published in the Proceedings in the National Academy of Sciences recently. And they looked at Chicago, LA, and New York and looked at, I think it was like three different violent crime categories: robbery, aggravated assault, and I think regular misdemeanor assault.


And what they found was that when you control for the massive decrease in the amount of time that individuals were actually spending in public, their risk of victimization per hour spent in a public space had actually gone up 30 percent, even though a lot of those crime categories in 2020 had seen a raw number decrease.


And if you look at cities like New York -- and there was an analysis done just last summer showing that foot traffic in midtown was still down 30 percent compared to 2019. Subway ridership was only 70 percent of what it was in 2019. The number of individuals in the office five days a week had just plummeted. It was way, way down. So I think a lot of cities have seen a lot of these routine activities’ trends stick.


And so what that indicates is that some of these numbers—particularly figures like robberies and even theft—don’t actually reflect how truly bad things have gotten to some degree because the numbers are depressed by the change in routine activities.


On the gun sales front, yeah. I mean, I think if you look at those numbers, again, the majority of the new sales went to already existing gun owners. And there doesn’t seem to be a very strong relationship in terms of the geographic analysis, right? Where the sales were occurring where not necessarily where you were seeing the increases.


One thing that we know from 2020 is that the crime spike did not reflect a broadening in terms of the overall concentration of gun violence in particular. It remained about as concentrated as it had been in prior years. David Weisberg, who was the inventor of the rule of crime concentration professor at George Mason, did an analysis for the Manhattan Institute in 2021, looking at 2010, 2015, and 2020 in New York City and found that crime— both violent crime and crime generally—was just as concentrated in each of those years. And, in fact, in 2020, the concentration had become slightly more tight. Yeah, I think that’s important to note.


In terms of the hypotheses, that makes sense. I mean, look. We saw an acceleration of already pre-existing trends, right? A preexisting trend in decarceration between 2010 and 2021, the prison population was down some 25 percent in this country at the state level, which is, again where about 9 out of every 10 prisoners are housed. The federal prison population has gone down significantly over that period. The jail population has decreased probably around 15

percent over that period of time.


The number of arrests being made by police officers has decreased about 25 percent over the last decade. The number of police officers on the street has decreased. We’ve got at least pretty good anecdotal or anecdata about the degree to which police officers feel a little more reticent in terms of their willingness to be proactive. And we do have decent data from prior analyses on what we might be able to expect in terms of effect on crime from those things.


I mean, one of the interesting analyses on that front was done by Roland Fryer and Tanaya Devi, looking at -- they were actually trying to assess the impact of federal pattern and practice investigations on crime. And what they found was that the pattern of practice investigations that were launched in the wake of a viral police incident were associated with very significant increases in homicides and crime overall. And the mechanism that seemed to be the most likely explanation for that effect was a reduction in proactivity, a reduction in stops, a reduction in traffic stops, a reduction in arrests done by the police officers in those five cities that they analyzed. So yeah.


I think when you just take a look at the system, take a look at the explosion, the electoral success of the progressive prosecutor movement, for example, which I think is fairly characterized by policies that adopt these blanket rules of non-prosecution for certain offenses and view prison as more of a last resort than a sort of standard sanction for criminal conduct: bail reforms in New York and Illinois and Chicago, etc.


New York has passed all kinds of reforms. In 2020, you had the bail reform, the discovery reform. In 2021, you had the parole reform. In 2018, you had the juvenile justice reform. You had the closure of Rikers Island. You had the elimination or the work around qualified immunity for the NYPD, the Right to Know Act, the diaphragm laws. And that’s just one city—not to mention the ten police reform bills signed into law in June of 2020 by Governor Cuomo, which did all kinds of things, including criminalizing the use of certain grappling techniques by police, which the city level also did.


And that’s a trend that we saw throughout the country. I think The New York Times did an analysis that in the year following George Floyd’s death. Thirty-five states had passed somewhere in the range of 140 different criminal justice reform bills, which is -- that’s a lot of movement in a very short period of time. And we haven’t necessarily seen that trend shift, although I do think we’re starting to see signs of that in certain states.


In Louisiana, we’re seeing some proposals to roll back some of the 2017 reforms there. In Tennessee, we just saw a proposed amendment to their constitution to expand pretrial detention. Last year, they passed a truth and sentencing regime. There’s three-strikes rule up for consideration.


But yeah. I think just overall, I think we saw a very clear trend toward decarceration and de-policing. I don’t think you can divorce that from the crime increase. And whether that continues is a different story. In terms of why we might be seeing a decrease now, I don’t know if we’ve ever seen five, six straight years of continued crime increases. It tends to go up, level off, come down, go up, level off, come down.


I think part of what we might be seeing in terms of the homicide decrease could be a function of the permanent incapacitation effect of the homicide increase of 2020 and ‘21. If you look at the profile of the typical homicide victim in most American cities, they’re pretty similar to the typical homicide offender.


And so to the extent that we saw a massive spike in homicides, we also saw a spike in the killings of individuals who were probably at high risk of committing a shooting themselves. How that might be showing up in the data is something, I think, that needs to be considered. So yeah.


There are a lot of factors that can go into explaining the fluctuation, but I do think that the broader trends of decarceration and de-policing probably have a good bit to do with those crime increases, which, again, I think are an extension of the increases that we saw in 2015 and ‘16. I don’t think it’s starting in 2020.


Vikrant P. Reddy:  Our host has returned and came off mute because we’re at time. Nevertheless, I’ll give you the last word, Stephanie, if you can keep that last word under 10 seconds.


Stephanie Kennedy:  Thank you for a spirited discussion, and I hope that you’ll go to our website. I realize I put the chat in a place where you may not have seen it. But the CCJ website, you can access our First Step Act analysis and go and look at our most recent crime trends report that Vikrant referenced and see how we’re tracking crime and the First Step Act over time. Yeah. And we look forward -- if you have questions that I didn’t answer here, feel free to email me.


Chayila Kleist:  Well, thank you. I really appreciate you all joining us today. It was a fantastic discussion that we could have gone on for another hour. Thank you also to our audience for joining and participating.


We welcome listener feedback by email at -- sorry, at [email protected]—new email. And as always, keep an eye on our website and your emails for announcements about other upcoming virtual events. With that, thank you all for joining us today. We are adjourned.