Probable Causation

View Original

Episode 8: Amanda Agan

Amanda Agan

Amanda Agan is an Assistant Professor of Economics at Rutgers University.

Date: July 23, 2019

A transcript of this episode is available here.


See this content in the original post

Episode Details:

In this episode, we discuss several studies on the effects of Ban the Box policies.


RESEARCH WE DISCUSS IN THIS EPISODE:


Transcript of this episode:

 

Jennifer [00:00:06] Hello and welcome to Probable Causation, a show about law, economics and crime. I'm your host, Jennifer Doleac of Texas A&M University, where I'm an Economics Professor and the Director of the Justice Tech Lab.

 

Jennifer [00:00:16] My guest this week is Amanda Agan. Amanda is an Assistant Professor of Economics at Rutgers University. Amanda, welcome to the show.

 

Amanda [00:00:24] Oh, thank you. Thank you, Jennifer, so much for having me. I'm very excited to be here.

 

Jennifer [00:00:27] So we're gonna do something a little different this week. Usually on the show, I talk with a guest about a specific paper that they wrote, stepping through what they did, how it contributed to the literature or policy implications and so on. But we decided to mix things up for this episode. So you and I both wrote papers on Ban the Box and have been heavily engaged in both the academic and policy conversations about this policy and are frequently asked to weigh in on the state of literature on this topic. And so we're going to have a conversation here today summarizing that literature and giving our perspectives on both what we know and the open questions that remain. And I know we agree about a lot of aspects of this literature, but I think we also disagree about some things. So hopefully this will be a fun chance to hash those disagreements out in a public setting for all to hear.

 

Amanda [00:01:12] Sounds good.

 

Jennifer [00:01:14] OK. Let's start with you introducing yourself. Could you tell us about your research expertise and the types of issues that you study?

 

Amanda [00:01:21] Sure. So I mainly study broadly the economics of crime, and I'm particularly interested in how criminal records interact with labor market outcomes and how various policies can impact the probability that somebody with a criminal record becomes employed. And that's kind of where I've been focusing for the last couple of years, really. And that's where Ban the Box kind of really started a lot of that and flows right into into that interest.

 

Jennifer [00:01:46] OK, so you ready to dive into these papers?

 

Amanda [00:01:49] Sure let's do it.

 

Jennifer [00:01:49] All right, so why don't you kick things off for us by summarizing what Ban the Box is for those listeners who don't follow this policy conversation as closely as we do.

 

Amanda [00:01:57] Sure. So Ban the Box is a policy that prevents employers from asking about criminal histories on job applications or in very early stages of the kind of job interview process. And so these policies are quite popular. They exist in about 35 states and 150 cities and counties as of right now. A majority of these policies are actually focused on public sector jobs. That is, the city or county has banned the box for its own hiring. And increasingly, states and jurisdictions are starting to pass these policies for private employers as well. So in 12 states and I think about 18 cities or counties, actually, basically any employer in that state cannot ask about criminal histories in the initial parts of the job application process. It is important to note, however, that the policies don't prevent eventual criminal background checks, right. So you can't ask the question on the job application. You usually can't ask it in a first interview. And they'll kind of legislate when in the process you can actually ask this question or perform a criminal background check. And that can vary from after the first interview to all the way until a conditional offer of employment has been made. But the employers always have the opportunity to eventually perform a criminal background check and potentially not hire the individual that that that is in question. And so the policies are really quite recent. I think Hawaii and Massachusetts were really kind of pioneers here. But most of these policies have come online in the last five or six years, let's say between 2013 and 2018.

 

Jennifer [00:03:30] Right. And so people like us heard about this policy and thought, oh, that could easily backfire due to something economists call statistical discrimination. And this is the idea that when you can't observe some characteristic that you care about, such as whether a job applicant has a criminal record, you try to infer that information based on other observable characteristics that you know are statistically correlated with the thing you wish you could see. So in this country, young, black and Hispanic men with less education are more likely than others to have a recent conviction that might worry an employer. And so if employers can't ask about who has a criminal record, they might try to guess and then might discriminate against young black men without college degrees, for instance. And we've seen this type of behavior in other contexts. So, for instance, Abby Wozniak has a paper on drug testing, where she finds that when employers are allowed to drug test their employees, employment of low skilled black men and black women increases by 7 to 30 percent. And this suggests that when drug testing isn't allowed, employers use race as a proxy for drug use and assume that black applicants are more likely to use drugs.

 

Jennifer [00:04:34] Similarly, Alex Bartik and Scott Nelson have a paper on credit check bans, and they find that preventing employers from checking the credit histories of job applicants reduces job finding by black applicants who have lower credit scores on average. And this suggests that when they can't directly check someone's credit, employers simply assume that black applicants have lower credit. So in other words, removing this information in effect broadens the discrimination to the entire group, obviously the opposite of policymakers' intent. And finally, there's an earlier literature on what happened when it became easier to access job applicants' criminal records in the first place due to the Internet and cheaper computer databases. And those studies generally find that having easier access to criminal record information increased employment for black men. So in the context of that broader literature, people like you and me immediately thought we should test for statistical discrimination when Ban the Box goes into effect. And you and Sonja Starr ran an amazing field experiment that has since been published in the Quarterly Journal of Economics. So tell us what you did and what you found in that paper.

 

Amanda [00:05:34] Yeah, so it was exactly in this context that we wanted to understand what was going to happen when Ban the Box policies went into effect. And were employers going to try to rely on other observable characteristics of job applicants to try to guess whether they had a criminal record, and could this have kind of unintended consequences? So in order to study this, we actually did what kind of is now called an audit study, where we sent a whole bunch of real job applications on behalf of fictitious applicants to low skill entry level kind of minimum wage type jobs in both New Jersey and New York City, both before and after private sector Ban the Box policies went into effect in those states. And so both New Jersey and New York City implemented private sector Ban the Box laws in 2015. Knowing that this was going to happen, we were able to sort of run this audit study both before and after to see how employer reactions to various observable characteristics might change once the employers were no longer able to see whether the applicant had a criminal record. And so basically with these fictitious applicants, they were all young men and we randomly varied whether the applicant was at least perceived by the employer to be white or black using racially distinct white and black names. We also varied whether the applicant had a criminal record, a felony conviction. We also, you know, in addition to race, we were also interested in whether employers might look at employment gaps as a potential signal that somebody may have spent time dealing with the criminal justice system or whether they had a GED versus a high school diploma as another potential signal that they might have had experienced some criminal justice contact or at least been more likely to. So we also were randomly varying those characteristics. And so we sent in the end, kind of 10- 15,000 of these applications, both before and after these Ban the Box policies went into effect, to sort of understand how employers were going to react to these characteristics. What we were tracking, our main outcome, was whether the employer called back our job applicant or emailed back, honestly. We were tracking both, but we kind of use callback as our generic term for that outcome, you know, basically whether the employer was giving a positive reaction to our applicants.

 

Amanda [00:07:50] So in general, the you know, something you know, somewhere between 10 and 15 percent of our job applicants received some sort of positive callback from an employer. If you look at those employers who had a criminal record question on their job application before Ban the Box goes into effect, for those employers, white applicants were only about seven percent more likely to receive a callback than black applicants in that era where they could actually ask this criminal record question. However, after the Ban the Box policy goes into effect - so now that criminal record question goes away and the employers can no longer tell whether the job applicant has a criminal record or not - we see that racial disparity in callbacks jump dramatically. So after Ban the Box and employers who had to remove that question from their application, white applicants were then 43 percent more likely to receive a callback than black applicants. That is a huge increase in racial disparities after Ban the Box goes into effect. And it really comes from sort of a combination of two things. Mainly, the white applicants in the post period after Ban the Box goes into effect, they were basically called back at the same rate as a white applicant that didn't have a criminal record in the pre-period. So it's as if kind of employers were looking at white applicants and saying, yeah, there's basically no shot that they have a criminal record, we're gonna call them back all as if they don't have a criminal record. However, in the post period for black applicants, one, we do see an increase in the callback rate for black applicants with a criminal record. So it does seem like black applicants with a criminal record have a higher probability of getting their foot in the door with employers. However, there is also a dramatic decrease in the probability a black applicant without a criminal record gets a callback. So those individuals are now no longer able to get their foot in the door with employers. And in fact, if we kind of take some back of the envelope calculations, it seems like that drop is swamping the potential increase in callback rates for individuals with criminal records and kind of lowering overall callback rates for black applicants. And so it does really seem like at least at this kind of call backstage that there is an unintended consequence of the policy of reducing potential employment opportunities for young black men without criminal records.

 

Jennifer [00:10:22] And the huge advantage of your experiment was that it was clean and randomized, which, of course, we love. The one downside was that you couldn't see who actually got jobs. And in this case, that might matter because the people with records who make it to the interview might later not be hired when their criminal records reveal during the background check. So your study, even with its already disheartening findings, might actually have made things look even rosier than they were. So my study with Ben Hansen in that in that scenario is a nice complement in that we have a less clean experiment, we exploit a natural experiment, but we can see who's actually employed. So in that study, we use data from the Current Population Survey, which we typically refer to as the CPS, one of our country's big national surveys that includes employment information as well as demographic information, although notably it does not include information on criminal records, as we know none of our surveys do. And we use the rollout of Ban the Box across places over time as a natural experiment. So we effectively compare employment trends for various demographic groups across places that adopt Ban the Box and places that haven't yet adopted Ban the Box. So if we think of Ban the Box as a shock to local labor market opportunities for certain groups, then we should see a change in employment after Ban the box goes into effect relative to places that adopt the policy at a different time. So it's essentially difference in difference design.

 

Jennifer [00:11:44] So what we find is that Ban the Box reduces employment for young black men without a college degree by five percent. We also find less robust evidence that it reduces employment for Hispanic men in the same group by three percent. We run a zillion robustness checks and additional tests as you do in these kinds of papers to complement that main analysis. And we find that the negative effects of Ban the Box are larger when unemployment rates are higher, that is when employers have more applicants to choose from, which gives them more power to discriminate. We also find that effects are driven by a drop in job finding for black men, rather than an increase in job loss, which is, of course, in line with what we'd expect if Ban the Box is affecting the hiring process. We can also look at who gets the jobs that young black men no longer get. So the story seems to be that public employers substitute from young black men to older black men who are a better bet if you are trying to avoid someone actively engaged in crime. Private employers substitute from young black men to young white men, which is in line with the results from your study where the context was private Ban the Box laws. So that's our paper. Our paper's now forthcoming at the Journal of Labor Economics.

 

Jennifer [00:12:51] What neither of our studies was really able to address directly was what happens to people with criminal records. So your study suggests that they got more callbacks. And I think both of us assumed going into this that Ban the Box would help some people with records. And the question we were really focused on was what was the net effect on black men given an increase in statistical discrimination against those without records? So more recently, there've been a whole bunch of new studies, some looking at employment, some looking at other things, but in particular, let's start out looking at the studies that really focus on the effect of Ban the Box on people with criminal records. So you're going to walk us through those studies. Let's start with the paper by Jackson and Zhao, two economists at the Boston Fed.

 

Amanda [00:13:34] Yeah, great. Yeah, and just almost to step back for just one second, just reiterate something that you said before is that this question is hard, right? Because our large national surveys, the CPS, the ACS, they don't ask whether somebody has a criminal record, right. And so it's hard, particularly in a broad national context, to really try to get at the question of, OK, well, how does Ban the Box actually impact the group it was meant to directly impact, which are people with criminal records who might be in the labor market. And so two of the papers that we're going to discuss just now kind of look at one jurisdiction and they were able to link administrative records, administrative criminal records with unemployment insurance records to really understand the labor market impacts on people with records, the people that were directly meant to be impacted by the policy, of Ban the Box policies in two different jurisdictions. And so that's great and really helpful and innovative that they were able to do this sort of data linking. And it's, you know, would be great if this sort of data linking could spread into larger jurisdictions and across the country.

 

Amanda [00:14:40] And so these two economists at the Boston Fed, as you just mentioned, Bo Zhao and Osborne Jackson were studying Massachusetts in particular, and Massachusetts' CORI reform, the Criminal Offender Record Information reform that went into effect in two parts. So in November 2010, basically, Massachusetts implemented a private sector Ban the Box policy. So they were amongst the earliest jurisdictions to implement a Ban the Box policy, in particular, one that was related to private employers. And so after November, private employers in Massachusetts could no longer ask whether somebody had a criminal record. There was actually an additional part of this CORI reform that went into effect in 2012, which actually affected how much information employers could get about individuals' criminal records. So they're going to focus, at least for thinking about Ban the Box, on that period between November 2010 and May 2012, when the additional record access law goes into effect, so that we can really try to isolate the impact of this private sector Ban the Box policy on employment for people with records. Now of course, the big problem, if we're looking at a single jurisdiction like one state, Massachusetts, and we only have data from within that one state, is to really try to understand what are the causal impacts of Ban the Box on employment for people with records. You know, can we find an appropriate control group, an appropriate counterfactual to try to understand if any changes that we're seeing are actually about the policy or about something else that may have been happening in early 2010 to employment trends for this particular group?

 

Amanda [00:16:21] And so what they did, which is kind of really interesting, is to think about- they have these- they were basically able to get access to the CORI records, to the Criminal Offender Record Information, the official criminal record database that Massachusetts has. And so that has everybody that's going to have some sort of criminal record in Massachusetts over a time period. I believe they have information from early 2010 up until sometime significantly past the policy. So what they look at is individuals who have a record in CORI before 2010. So that is that those individuals, as of the time, November 4th, 2010, when Ban the Box goes into effect, they are going to be potentially affected by Ban the Box. They have a criminal record, now employers can't ask about this information. Now in their dataset, they also have information on individuals who are not going to have a criminal record until much later. So let's say after 2012, two years past Ban the Box, they have information on those individuals. But as of 2010, when Ban the Box goes into effect, they don't have a record. Right. So they shouldn't be directly affected by Ban the Box. And so they're going to use those individuals who have- don't have a criminal record until after 2012 basically as a control group to try to say, well, those individuals are somewhat similar. They're people who are going to be active in in- have active criminal justice contact sometime during this time period, but shouldn't have been directly affected by Ban the Box when it went to it went into effect in 2010.

 

Amanda [00:17:53] Now there's going to be various issues. It's difficult, natural experiments, trying to come up with a control group, trying to come up with appropriate counterfactuals. Right. One major problem here is there's gonna be a huge age imbalance between these two groups. We have a set who aren't going to be actually committing their crimes until after 2012. We have a set that committed them before 2010. Those groups clearly have large differences in terms of their age. So the authors are going to do various things to try to deal with the covariate imbalances between their two groups. And across many different specifications that they try many different types of dealing with this kind of covariate imbalance, what they actually find is that after November 4th, 2010, there are significant reductions in employment for people with records, the treated group, versus those people that aren't going to have records until after 2012 after Ban the Box goes into effect. Which is in some sense fascinating and puzzling, right. We might have hoped that Ban the Box was potentially going to increase employment for people with records since hopefully they can now get their foot in the door with employers. And maybe even though they're going to do that eventual background check, maybe they're still going to go ahead and hire that individual because they've already realized they're a good fit. Or maybe we were going to think it had no effect on employment because employers could eventually do that background check and maybe they were still just going to not hire the same individuals they wouldn't have hired in the first place once they see that criminal background. What I think we might not have expected ex ante was that it was actually going to reduce employment with this particular group.

 

Amanda [00:19:35] And so I think it's a really interesting question to then think about why this is happening. The authors, you know, there's there's not much more they can do to explore this, I think within the context of their current data. They posit that by, you know, kind of removing the criminal record question from these applications, an applicant with a criminal record is now seeing that their- doors are opening up for them. There are lots of opportunities out there, they might be very hopeful, and they may actually increase their reservation wages. And so that this could be kind of a labor supply effect of sorts. I don't know, you know, maybe maybe that could be happening. You know, kind of always wondered - and I think maybe you and I have discussed this in the past, too - you know, whether this could be about discouragement, if they're going out there. And let's let's say, you know, the results of my study, you know, would totally transition to Massachusetts as well. And that there's a bunch of people who are now getting their foot in the door, that're actually getting a callback. But then when it comes to the end of the application stage, the employers eventually doing that background check and they're not getting the job. So now they're, you know, they're hopeful, but then they're not actually getting jobs. And that kind of discourages them and they step back from the labor market. That seems like one possibility. Of course, you know, of course, we can also question things about the estimation strategy, whether this is appropriate counterfactual. Those are kind of really less fun to try to try to posit about. And so I think this is this is a really interesting piece of the puzzle that still leaves these open questions at the end. I don't know. What have you thought about that reduction that they find?

 

Jennifer [00:21:16] Yeah, I totally agree, it's fascinating. Yeah. My best guess is that it's something along the discouragement lines, that, you know, this- Ban the Box effectively removes information from the hiring process. And, you know, the hiring process is essentially a matching problem. And so if you remove information, it makes matches harder to make. And so maybe everyone now is wasting their time interviewing for jobs that they don't have any chance of getting. And both that, you know, that that waste of time is going to distract them from getting the jobs they actually would have gotten gotten and also might discourage them along the way. So I think I think the result is consistent with that story. And the other fascinating piece is they do have a companion paper that finds that even though they find employment falls in this paper, they have another paper, where they show that recidivism also seems to fall, which is again puzzling. We would think the two would go in opposite directions. And so maybe, you know, maybe Ban the Box is providing some hope and that's keeping people out of criminal activity even though they're not getting jobs. I don't know. It does raise as many questions as it answers. And I agree, there is, of course, the possibility that their control group is just not a good counterfactual after all. But I agree that it's also a less fun possibility.

 

Amanda [00:22:30] Yeah. So, yeah, definitely. Definitely a fascinating paper.

 

Jennifer [00:22:34] So next up is a paper by Evan Rose, who uses data from Washington State. So walk us through that paper.

 

Amanda [00:22:40] Yep. So this is going to be very similar in spirit where we're going to be able to link- he's going to be able to link administrative criminal record data to UI data. So we can really track employment outcomes for individuals with criminal records. This time, the context is the state of Washington, where Seattle implemented a Ban the Box policy for private sector jobs on November 1st, 2013. And so, again, then the question is, how do we come up with a counterfactual in terms of doing this before and after analysis? What works for, you know, this this paper that might not have worked for the past paper where the entire state implemented the policy, here he has data from all of Washington State, but only Seattle implements the policy. And so basically throw out the paper in terms of thinking about the impact of Ban the Box on employment outcomes. He's going to use either neighboring counties or Spokane, Washington, which is much farther away, but also a major city in Washington as basically control groups. And so we're going to look at what's going to happen to employment for specific groups of people with records after Ban the Box goes into effect in Seattle versus what happens to trends in employment for people with records in either neighboring counties or in Spokane, Washington.

 

Amanda [00:23:56] And so what what he finds across various definitions of people with records, whether we're looking at people on probation or people who've been recently released or or kind of various definitions is basically no change in employment for King County, where Seattle is, versus these other potential control groups after Ban the Box goes into effect. When we look at this, it's not like we're seeing these large increases in either employment or earnings in Seattle versus these other counties. You look at the trends and they look very similar beforehand and they look very similar afterwards. And the basic conclusion here is that Ban the Box, the kind of private sector Ban the Box that Seattle implemented, doesn't seem to be increasing employment for people with records. Now, it's not decreasing it the way the previous paper in Massachusetts found. We're basically finding these these kind of null effects. There's nothing happening in terms of employment for people with records, which, again, we may have expected if what is happening is that employers are eventually doing these background checks and still not hiring the individuals who may potentially be getting their foot in the door with the advent of these Ban the Box policies, but not actually eventually getting hired. Now, why exactly, you know, this paper is finding no effect and the previous paper's finding a negative effect - that, you know, we might be less clear on, it could be a jurisdictional thing that just Massachusetts is going to have different policies and different reaction than Seattle. Could be about time period, right. We're talking about 2013 versus 2010. Could be about the estimation strategy. So neither paper can actually exactly replicate the estimation strategy of the other due to just the way the data and the policy is set up. And so, you know, it could be about the counterfactuals or control groups, lots of open questions. But these two papers combined, right, are not giving us much hope that private sector Ban the Box laws are increasing employment for people with records.

 

Jennifer [00:25:55] Right. And just to clarify, I'm pretty sure that the Seattle policy- it is a private Ban the Box law, but it also included Ban the Box applied to public employers, right.

 

Amanda [00:26:02] Oh, yeah, yeah, yeah. To be clear, yeah, most of the ones that have private sector Ban the Box laws are also pertaining to public employers as well. It's it's usually kind of on top of a public employer law.

 

Jennifer [00:26:11] Right. So his results are also also relevant to those public Ban the Box laws. Yeah, I think this is a really nice paper. And as you said earlier, like, you know, this is the data we wish we had at the national level. And so this is this this is the kind of study we wish we could just do nationwide. And so I think the big question, like I'm left with, after this paper- because he really I think in many ways, it's a much cleaner experiment. He's able to sort of look at other counties as the control group, and he doesn't have to do sort of the propensity score matching that the Fed guys had to do in Massachusetts and to to balance ages and stuff, like he has really nice control groups. And so so I'm I'm convinced that the results are internally valid. I think the big question is to what extent Seattle is representative of other places. And that's, of course, you know, we won't know unless we do the studies in other places. There's no immediate reason that comes to mind that I would expect these results not to apply to other places. So I think it's a nice study.

 

Jennifer [00:27:10] So next up, there's a paper by Terry-Ann Craigie, who attempts to get at this question using data from the NLSY. So tell us about that.

 

Amanda [00:27:17] Yes. So in this paper, it's really focused on public sector Ban the Box policies. And she's going to use data from a nationally representative survey, which you mentioned, the NLSY 97, the National Longitudinal Survey of Youth from 1997, which is a representative sample of youth in the United States who are between the ages of 12 and 16 in 1997. And basically this survey asks them questions every year about their schooling, about their employment, also self reported information about their criminal history. Have they been arrested? Have they been convicted? Are they married? Just a whole bunch of questions have been asked of these 9,000 individuals over the last 17 years or so.

 

Amanda [00:27:58] So the latest round of data is available in 2013 where these individuals are between the ages of 29 and 34. And, you know, they come from across the United States. And so she's able to use kind of, you know, a more nationwide measure here where she's going to use the difference in difference style, exploiting the fact that public sector Ban the Box policies rolled out at different times. So very similar to kind of what you and Ben were doing in your paper. And then rely on these kind of, you know, admittedly self reported conviction status measures. And the NLSY also tracks whether somebody is employed in the public sector or not. So the question she really wants to ask is when public sector Ban the Box policies go into effect, how does that impact whether somebody with a conviction or self reported conviction is changing their probability of being employed in the public sector? And in addition to that difference in difference, she also adds what we would call a triple difference. And additionally, differences this for convicted individuals versus non convicted individuals, because basically what she's finding is that there are not parallel trends. The trends in public sector employment are not looking very similar before Ban the Box goes into effect for banned states versus non banned states. So she adds this additional difference between those who report having a conviction and those who report not having a conviction. And looks at the differences in employment for those two groups in banned versus non banned states before and after Ban the Box policies go into effect.

 

Amanda [00:29:33] And what she's actually finding is an increase in employment in public sector employment for convicted versus non convicted individuals in banned versus non banned states after Ban the Box policies go into effect. Always very fun to try to explain a triple difference in an easy sentence. And in addition, she also- she claims to find no evidence of statistical discrimination, as in basically finding no significant decrease in public sector employment for low skill men of color, both black and Hispanic men, after ban the box policies goes into effect. In some sense, I think trying to replicate some of what you and Ben did in the CPS in this kind of smaller NLSY dataset. Although I think that's where we can quibble a little bit about whether there's no evidence of statistical discrimination or we just don't have power to estimate this evidence of statistical discrimination. There are pretty wide confidence intervals on these these estimates of potential statistical discrimination. And so that is kind of the upshot of the paper in terms of what she's finding. I'd be really interested to hear what you think about those effects kind of squaring with some of your research and some of our past research.

 

Jennifer [00:30:48] Yeah, I think first I want to back up a little bit and just think about this question about like why we might expect Ban the Box to have different effects in the public sector than the private sector. So I think it totally seems reasonable to me that government government employees might care less about criminal records to begin with. Right. And so this this you know, this raises the question of, like, why employers care about these criminal records. And one reason, at least employers say, they're worried about legal liability. So it's totally possible the government is just not subject to the same legal liability concerns that private employers have. And and so but that's a reason that Ban the Box would have no effect. Right. We just wouldn't expect discrimination against people with criminal records to begin with-.

 

Amanda [00:31:30] In the first place.

 

Jennifer [00:31:31] In the first place, right. And so then you you take that information away and then you wouldn't expect you know, I would expect that not to matter then if they didn't care in the first place. It's not a reason we'd expect an increase in employment unless the main effect of Ban the Box is to pull more people into the applicant pool, which is definitely an effect it could have had. But that's not the effect of removing the criminal record question. That's the effect of advertising that you're willing to hire people with records. So this is where we kind of get into like an issue of like, what exactly are we testing here? Anyway, that's just a little bit of background on how I think about this question of do we expect different effects for public employers?

 

Jennifer [00:32:05] So as you said, I mean, so Terry-Ann does try to replicate a little bit of what Ben and I do. We also have information on whether people are employed in the public or private sector and find reductions in both places. So I think of- the reason we use CPS data is that we think those are better data to get at these employment questions. I mean, the reason to use NLSY data is it's the only nationally representative data that we have that has anything like criminal history data in it. And so if you want to study this question, this is the dataset that you need to use. The problem is then it just becomes a really small sample. And it is self-reported data and there is some evidence that people do change the way they report their criminal history, depending on various circumstances. And so one concern I have about using this dataset for a study like this is that when Ban the Box goes into effect, that reduces the stigma associated with having a criminal record, hopefully. That's the goal anyway. And then so that could actually lead some people who are otherwise employed and have their life together to admit to the surveyor that they do have a criminal record. And so it could change the composition of people that are in these different samples. So it just makes it really complicated in addition to this small sample and the wide confidence intervals and everything. I think it makes it really tough to interpret what's going on here.

 

Amanda [00:33:18] Yes, that's so interesting. I'm not sure I thought very hard about the fact that- I knew, I mean, we always think about self-reported criminal records, but exactly what this- Ban the Box could actually kind of incentivize people to answer those questions a little bit differently, I hadn't I hadn't thought that I haven't really thought hard about that that part of things, which which we don't have administrative data, of course, because the administrative data is just the administrative data. So that is that is really interesting. Yeah.

 

Jennifer [00:33:43] Yeah. So there is also a paper by Danny Shoag and Stan Veuger, so this is kind of moving on from looking at these direct effects on individuals with criminal records, but they're still kind of trying to get at this question. So their their paper focuses on neighborhoods. So basically, they're using a subset of cities where they were able to obtain neighborhood level crime data from the year 2000. And they identify neighborhoods that were high or low crime in that year, the year 2000. Then, compare the number of people employed in those neighborhoods over a decade later, and they compare places with and without Ban the Box. And they find what I guess I think of as suggestive evidence that Ban the Box helped high crime neighborhoods relative to low crime neighborhoods. And where I get hung up on this paper, or would I get hung up, hung up on with this paper is that they don't control for the likely changing composition of those neighborhoods over time. So I'm sure we can all think of places in cities we know where neighborhoods that were pretty rough in 2000 are pretty nice now. So I know D.C. pretty well, 14th Street in D.C., for instance, 14th Street. You know, if 14th Street includes more people with jobs now than it did in 2000, I'm not sure what that tells us exactly. It's just, you know, it's gentrified heavily in that interim period. So anyway, there might be something here. But the paper doesn't include enough information or checks for me to to really understand kind of the results and put them in perspective with the other papers we've got. What's your take?

 

Amanda [00:35:11] Sure. I totally I think I think that has always been one question I've had with that paper. And unfortunately, you know, it's kind of a data issue. We don't have a lot of information on to be able to really pinpoint time trends in kind of high crime or low crime neighborhoods like that. That's a really interesting kind of dataset that they were using to try to do that. But it's not a dataset that's been repeated since 2000 in a way that we could really get at those time trends. And so I think, you know, the authors likely know that this is going to be a limitation of what they're doing. And it's it's really too bad that we don't have those kind of time trends to really pinpoint and see what's happening just before and just after Ban the Box in terms of employment.

 

Amanda [00:35:54] And that paper also has some- uses data from the American Community Survey, if I recall correctly, to try to look kind of generally at employment trends before and after Ban the Box again in kind of a similar style as what you did with Ben. But they don't do it kind of in an age differentiated way. You you both were really looking at young, low skill black and Hispanic men. This paper looks at kind of overall employment trends for black men, black women. Hispanic men and Hispanic women, actually now, I don't have the paper right in front of me, I can't remember. And I always thought it was interesting because I think they're finding increases in employment for black men overall in the ACS after Ban the Box goes into effect, which originally people said was contra what you and Ben were finding. But even thinking back to the discussion that we had just 20 minutes ago, you said that it was the kind of- some of the substitution that was happening was from younger black men to older black men. Black men who may now not be actively engaged in criminal activity. And that would totally kind of reconcile these results if we were able to look at the different different age groups in the ACS analysis that Shoag and Veuger are doing versus kind of the CPS analysis that you all did. And so I just remember that from a couple years ago, I think there was always this like that, that there is this tension between these two papers but it actually didn't feel like there was a lot of tension. They're kind of telling the same story, maybe not quite as nuanced with the ACS data, but it seems like it's kind of all pushing in the in the same direction.

 

Jennifer [00:37:31] Yeah. Yeah. We wound up adding a robustness check in ours where we use these ACS data because they, you know, they just look like there there might be a different effect. And we wound up finding- I mean, the ACS data, there was some- they like changed the way they asked about employment in 2008, which was really inconvenient. And so there's sort of there's sort of this additional question about like whether that question can actually be useful going over time, and anyway. So so I think that's part of an issue here, which we talk a little bit about in the paper. But but overall, basically find the same things. It's just noisier. I think the other breakdown they don't do is by education group. And so yeah, so we were really focused on like, OK, who- which- what is the group that is most likely to be helped or hurt by Ban the Box, if it helped, if the policy works and hurt if employers are statistically discriminating. And they were focused on this question of like, what is the- what happens to overall black employment, which is perhaps an interesting question, but I think not as interesting in this conversation about statistical discrimination. And so, yeah, I think part of it is just substitution between groups and questions then about like, well, are we okay with jobs going from young black men to older black men? Yeah.

 

Amanda [00:38:41] And this paper, Danny and Stan's paper, they also- this this kind of increase in employment for high crime neighborhoods to the extent that we want to call them high crime neighborhoods - so putting aside the trend question for just a second - it does happen mainly in private- or public sector employment. Right. So as going going back to thinking about this question about public versus private sector, whether there might be a difference, you know, we might put this all together in terms of thinking about the exact question that we're asking and the kind of let's call it the second part of this podcast, which was what is the actual impact of employment for people with records? Right. And it seems that we have evidence of an overall drop in states that have both public and private Ban the Box policies in terms of employment drop or no change. Maybe some suggestive evidence that has its question- limitations for the potential for some increases in employment. For the specific group of people with records in the public sector less clear, like it seems like it's clear that the kind of overall impact may be null or negative for the whole group of young, low skilled black men in terms of their employment in the public sector, but thinking about this specific group, potentially suggestive evidence of an increase in employment for people with records in the public sector. Is that how you would kind of read and reconcile these four papers in terms of their contribution to this question about employment for people with records?

 

Jennifer [00:40:12] I think your read's a bit more generous than mine is. These papers that find different results, that suggest benefits in the public sector have important limitations, as we've discussed. And my read is that those limitations are just as likely to explain the different effects that they're finding. These are both working papers. So they might be updated to address some of these issues. But at this point, my questions about whether those papers are isolating the causal effect of the policy change make me skeptical that one, Ban the Box is having beneficial effects on public sector employment. And two, that public Ban the Box laws avoid the unintended consequences that we highlighted earlier. I think we simply have better evidence at this point on these questions from the other papers we talked about so far, particularly the Rose paper. And in terms of the effects on people with criminal records and on the unintended consequences question, my paper using the much larger CPS dataset.

 

Jennifer [00:41:04] So let's move on to another paper that is interested in the effects of Ban the Box on crime. So this is a new working paper by Joe Sabia and colleagues that finds that Ban the Box appears to increase crime rates, especially for Hispanic men, not for black men. Their analysis suggests that this is due to differences in social service receipt. And that's sort of the angle of this that I just find fascinating. So they use the NLSY and the American Community Survey. They see drops in employment for both groups in both datasets, which is sort of interesting in contrast to the the papers we were just talking about. But I think the the piece that they add to this, which is fascinating, is they then see an uptick in receipt of food stamps and similar safety net programs for black men after Ban the Box goes into effect. But no similar change for Hispanic men. So they argue that safety net access helps mitigate the effect of job loss for black men. But that job loss plus limited access to safety net programs leads to more crime committed by Hispanic men. So this is a, you know, new, more recent working paper. It's still going through the rounds and peer review and all that. What's your take on this paper?

 

Amanda [00:42:16] Yeah, no, I think I think this is really interesting. I'm not I'm not sure I would have a priori guessed that this was going to interact with social service receipt in this way. And it's so interesting to think about how these other policies can also interact with Ban the Box in a way that kind of once you hear it, it's like, oh, yeah, that makes some sense. Right. Which is which is kind of the fun of doing empirical research in this way. You know, so I think that's a really interesting paper and I'm fascinated to see where it goes.

 

Jennifer [00:42:45] Yeah. OK. And finally, speaking of policies that interact with Ban the Box, there's some other evidence in two separate papers that Ban the Box leads job applicants and employers to favor occupational licenses that are unavailable to people with criminal records. So the first paper's by Peter Blair and Bobby Chung, the second paper is by Riccardo Marchingiglio. I hope I'm saying his name right. What seems to be going on there is that when Ban the Box reduces information available to employers about applicants' criminal records, employers find other ways to get that information. Or alternatively, job applicants find other ways to signal that information. Riccardo has a great line in his paper about how his results show that workers are willing to pay a substantial cost, that is obtain an occupational license, to buy back the box that policymakers banned. I think all this speaks to the issue of whether employers view criminal records as providing valuable information in the hiring process. And they they seem to do so. And to me, this highlights the fact that removing information from a market like this can have unintended consequences if we're not fully aware of how people are using that information and how they will adapt if we take that information away. Anything to add about those occupational licensing papers?

 

Amanda [00:43:57] Absolutely. And just to clarify, the reason that these occupational licensing licenses are a good signal is that most occupational licenses, you can't get them if you have some sort of felony conviction. Right. So they in and of themselves can be a signal that somebody doesn't have a conviction. And so this kind of goes right into a lot of what we were saying at the beginning about statistical discrimination and employers trying to get around the lack of information that they desire. And interestingly, as you already mentioned, the applicants themselves trying to signal this information. And so that's why that's why it works. And it's a, again, kind of really interesting additional policy to think about in this context.

 

Jennifer [00:44:37] OK. So that's the state of the literature. So let's talk more about what it all means. So when policymakers come to you and ask what they should do about Ban the Box, what do you tell them?

 

Amanda [00:44:52] This is a really interesting question because my answer has definitely changed over time. I think just after, you know, your and my our - mine and Sonja Starr's paper and your and Ben's paper came out - I was quite cautious. Right. Because what we had shown was that there was a cost for young minority men without criminal records. It was reducing their employment opportunities. But I always said, OK, but there's there's always tradeoffs in policy. And then the question was going to have to be, does this come with also an increase in employment for people with records? And we might, as policymakers or society, be willing to trade off this reduced employment opportunities for young minority men without records for increased opportunities for people with records because particularly because that increased employment could have positive externalities in terms of the reduction of future criminality, the reduction of future recidivism, the reduction of future incarceration. And so I was always saying, look, I I'm not ready to say that we shouldn't necessarily be implementing this policy because it could have these these positive impacts.

 

Amanda [00:45:56] However, as the literature has evolved, right, and particularly thinking about the Rose paper that we discussed and the Zhao and Jackson paper that we discussed, there isn't really robust evidence that Ban the Box policies are doing that positive thing that we wanted them to do, are actually increasing significantly employment for people with records. And so now we have a cost on one side, we have this reduced opportunity for young men of color with no records, without the subsequent increase in opportunity for people with records. And so then that leaves me going, I don't know, you know. We probably need to be seeking out other policies that are going to help increase employment opportunities for people with records without this kind of subsequent harm for non criminal justice involved individuals. Of course, the question that the policymakers and you are probably going to ask me is what are those policies? And I think that's where, you know, we have a lot on the research frontier to really think about what should those policies be and what should they look like?

 

Jennifer [00:46:59] Yeah. And I think, you know, talking about alternative policies is is really the key here. I think a lot of people, when they talk about Ban the Box, they think about- they think of this conversation as being it's Ban the Box or nothing. It's Ban the Box or you don't want to help people with criminal records. Right. And and so I- when I talk to policymakers, I try to reframe the question as it's Ban the Box or what? Or what else? And I think, you know, my takeaway from from all these papers is that removing information about someone's criminal record doesn't seem to help people, you know, with with criminal records and doesn't and seems to actively hurt people without criminal records because employers do care about these criminal records for some reason. Right. And figuring out what it is that employers are worried about and why they care so much about these criminal records is the key then to think to to finding ways to address their concerns so that they are more open to giving people with the record a chance.

 

Jennifer [00:47:58] And so, you know, I put kind of the possible avenues forward here into three buckets. So one is to help people with criminal records find other ways to signal their work readiness. So if employers think a criminal record is a signal that you're just not a good employee, you're not gonna show up on time, maybe you're not going to be sober or whatever else, find ways for people to, you know, go through rehabilitation programs or get recommendations from supervisors or something like that to signal that they'd be a good employee. Another possibility is to find ways to shift the legal liability from employers to courts or government. And so, you know, if employers just worry that, you know, if they hire someone with a record and that person commits a crime on the job, they are going to, you know, be sued or maybe the bad press will put them out of business. If we could have the courts guarantee someone or something like that instead, then that could shift the risk in a way that could be helpful. And the third is just to invest in rehabilitation, which we have, which is the hard thing to do and the expensive thing to do. But frankly, we have not done much of it in this country. And so we need to, you know, recognize that there are real challenges that this population faces and maybe the employer's concerns are legit in a lot of situations. And we need to help people become better employees. And this is just to kind of put a policy alternative on the table that seems to have some promise.

 

Jennifer [00:49:27] There have been a couple of nice studies, audit studies by Peter Leasure and colleagues looking at the impact of rehabilitation certificates, which are court issued certificates that someone with a record can get. They can go before a judge, talk about, you know, the job training program they completed, that you know- present drug test results, whatever it is, have character witnesses, anything they want. If the judge is convinced that they've been rehabilitated, they can give them this piece of paper that then they can give to an employer or a landlord. And so these audit studies show that when someone has one of these certificates, it essentially completely wipes out the effect of the record in terms of getting callbacks and and positive responses to housing inquiries. So this really seems to help people with a record get their foot in the door in the way that Ban the Box was hoping to. And it does that by providing more information rather than taking information away. So I think it just provides you know, we don't have to do this, but I think it provides a really nice example of a creative policy that by, you know, perhaps it's addressing the legal liability concern, perhaps it's addressing the signaling concern. But, we find another way to to help this population, that that actually seems to work.

 

Amanda [00:50:47] Yeah, absolutely. You know, I think those those certificate things are very interesting. And if I recall correctly, the certificates do exactly as you just said, both both limit employer liability for negligent hiring, you know, and obviously in some sense provide a signal of work readiness and a potential signal about productivity. You know, it sort of of course would be interesting also to try to disentangle those two pieces and see, you know, where exactly we need to push in terms of a bigger push. But I think those are to me really fascinating.

 

Jennifer [00:51:16] Yeah, yeah. So that is definitely, you know, one one avenue of a potential- you know, we think about research frontier here, that's one one area we need more work and teasing those those issues apart would be fascinating. What are some other big questions in your mind that need to be answered?

 

Amanda [00:51:31] Yeah. So you totally took my main one, which is why exactly are employers reluctant to hire people with records so we can really push on that policy angle. You know, I think one thing that I've been interested in in terms of potential alternative policies is things like expungement and sealing and these sort of automatic records sealing sort of acts that various states are starting to pass. These seem to have become also pretty popular to ease the availability of expungement and sealing or make it easier or make it more automatic. And then, you know, what this does is kind of one step beyond Ban the Box, because now the employer will never be able to see this criminal history information. And in fact, the individuals can report that they basically don't have a criminal record assuming and of course, the expungement took away the entirety of their record and not just part of it. However, usually when we're talking about expungement or sealing, we're often talking about relatively minor crimes that happened a long time ago, more than five, sometimes more than ten years ago. In limited cases in various states, there are things that you can expunge that are more recent or that you can seal that are more recent.

 

Amanda [00:52:40] So it's interesting to think about what impacts this particular policy can have. There's a question about whether it could have similar kind of Ban the Box style unintended consequences. I think that there's just less avenue for that here. So the idea would be, you know, the employers know that there's easier expungement now. And so knowing that, they may be, you know, less likely to call back a young black or Hispanic applicant thinking that they might have one of these expunged records. That is entirely possible, of course. I think Ban the Box was really salient. Right. A hiring manager used to have an application that had a question on it. And now they don't. They see very clearly that they're not seeing this information. Right. The idea that they kind of are really, really aware of expungement policy and might start to statistically discriminate in light of that expungement policy, knowing that now they can get rid of these ten year old records is plausible, but less likely to me. So I think it may not have those sorts of unintended consequences. What's interesting then to know is, does it have the intended consequences? You know, are the individuals with records, you know, that are ten years old, have they been kind of out of the labor market too long for this to help them? Are they going to respond to the expungement by actually increasing their job applications? Kind of what- I think it's just interesting to think about what are the potential impacts of these sorts of expansion of expungement and sealing as kind of a different way of, quote unquote "hiding criminal records" in a way that may have less of these sorts of unintended consequences? But I don't know.

 

Jennifer [00:54:17] Yeah, I guess I think of those unintended consequences as being, I guess, more plausible than than you do. And so, you know, in my mind, this is why we need the research. Right. And I believe you're working on this, so I'm looking forward to seeing what you come up with. And this is also where the avenue through which those rehabilitation certificates were working really matters. Like if it's about- if it's all about the legal risk, if it's all about reducing legal risk, then expungement could be really helpful because you can't you can't use an expunged record as the basis for a lawsuit or bad press or whatever. If if it's about the signaling, then we we could more likely be in this scenario where employers, you know, are really just trying to avoid hiring someone that they think is engaged in crime. And they now know the absence of a record doesn't mean that there was no record anymore. And so it could backfire if that's if that's what they're worried about and that's the way they're using them. So, yeah, so we need we need the research.

 

Amanda [00:55:14] Yeah, definitely. And I guess I mean, we we can say we can leave it at there's also this just much bigger question that kind of comes in the heels of the way the advocacy community reacted to some of our our own research, you know, which was that this is basically racial discrimination. And what we need to do is eliminate that racial discrimination from the labor market and not necessarily kind of reduce Ban the Box policies. And so maybe there's a bigger just a really, really, really big question about how we can reduce bias and stereotyping in the labor market in general, but I don't think that there is a easy way forward in terms of thinking about that. But that's one of those kind of big, big picture questions.

 

Jennifer [00:55:58] Yes. And for the record, we would both, I think, love to write the papers about how we eliminate racial discrimination. So if anyone has ideas, we would just gladly, we'd love to do that. And so since people do have those objections to our, or at least my, calls for for stopping passing Ban the Box policies and potentially repealing Ban the Box policies is the results we found were the effects of these policies under kind of current enforcement of racial discrimination laws. If we think that there are dramatic ways that we could increase enforcement of those laws, then perhaps the effects of Ban the Box would be different. But-.

 

Amanda [00:56:35] We need research.

 

Jennifer [00:56:36] Yeah, yeah. I guess we need the research. It is also I am more skeptical than most lawyers, I guess, that we can enforce our way out of this. But but definitely an important research area as well. I think, so I will also kind of add to your your list, I think just in general we need more work on what works to facilitate the successful reintegration of people with criminal records into their communities. I think it is pretty ridiculous that as a society, we've spent decades locking people up with no investment in education or training or rehabilitation. And then we blame employers for not wanting to hire them when they get out of prison. I think, you know, a large share of this group has real challenges that they need to overcome. And as a society, we need to do better in helping them. So I spent the last couple years reading lots of papers and writing a review on how to reduce recidivism. The most surprising thing to me, actually, was that employment doesn't seem to be as important as I think we all think it is, though money is important, which, you know, may be not surprising. But other types of interventions that seem useful are mental health and substance abuse treatment. It's remarkable how little we know, especially about the latter. So I guess my summary view is that we need to stop wasting our time with Ban the Box and focus on experimenting with other things. How about you? Any any final thoughts?

 

Amanda [00:57:56] No, I think that that would definitely be great if we could do a little bit more experimentation and understanding on all of these questions. Yeah.

 

Jennifer [00:58:05] My guest today has been Amanda Agan from Rutger's University. Amanda, this was so much fun. Thank you for being here.

 

Amanda [00:58:11] Thank you so much for having me.

 

Jennifer [00:58:17] You can find links to all the research we discussed today on our website, probablecausation.com. You can also subscribe to the show there or wherever you get your podcasts to make sure you don't miss a single episode. Big thanks to Emergent Ventures for supporting the show. Our sound engineer is Caroline Hockenbury. Our music is by Werner and our logo is designed by Carrie Throckmorton. Thanks for listening and I'll talk to you in two weeks.