Probable Causation

View Original

Episode 52: Katherine Eriksson

Katherine Eriksson

Katherine Eriksson is an Associate Professor of Economics at the University of California at Davis.

Date: June 22, 2021

Bonus segment on Professor Eriksson’s career path and life as a researcher.

A transcript of this episode is available here.


See this content in the original post

Episode Details:

In this episode, we discuss Prof. Eriksson's work on how access to education affected incarceration in the early 20th century:

"Education and Incarceration in the Jim Crow South: Evidence from Rosenwald Schools" by Katherine Eriksson.


OTHER RESEARCH WE DISCUSS IN THIS EPISODE:


Transcript of this episode:

 

Jennifer [00:00:08] 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 [00:00:18] My guest this week is Katherine Ericksson. By the time you're listening to this, Katherine will be an Associate Professor of Economics at the University of California at Davis. Katherine, welcome to the show.

 

Katherine [00:00:29] Hi Jen, thanks for having me.

 

Jennifer [00:00:30] So today we're going to talk about your research on how access to education affected Black incarceration rates in the early 20th century. But before we get into that, could you tell us about your research expertise and how you became interested in this topic?

 

Katherine [00:00:43] Yeah, sure. So I'm an economic historian, so sort of broadly work on US history, mostly topics that we would sort of think of as labor economics. So education's a big interest of mine. Incarceration obviously has been as well. I've done a lot of work on immigration, migration, health, so like just a general sort of labor market topics in early 20th century US history. How did I get to this topic? This was my job market paper way back when in 2013. So I started this paper probably in 2011, which tells you a lot about publishing lags. Yeah, it's terrible. And I don't know, it came out of I was really interested always in sort of the social returns to education. And I think that literature, even as an undergrad, I was like, this is really cool thinking about, you know, not just years of schooling and wages, but, you know, health, crime, other different things. And this paper was kind of almost an accident. I was working on a paper in South Africa actually looking at a policy change in education and juvenile crime. And that paper sort of killed itself when I found some pre-trends. I found some earlier data and my identification and everything fell apart. So one of my friends actually- I think one of your grad school colleagues, Roy Mill actually said to me, oh, you should think about Rosenwald schools. And so I was like, OK cool. So I sort of ended up doing this paper that sort of put me in the US history field instead of development. So it's kind of funny how the world- just random things happen and that determines your career.

 

Jennifer [00:02:08] Different path. Excellent. Yeah, so your paper is titled, "Education and Incarceration in the Jim Crow South." It was published in 2020 in the Journal of Human Resources. So in this paper, you focus on the effects of the construction of these so-called Rosenwald schools. So what were these schools and why were they built?

 

Katherine [00:02:26] Yeah, so they're called Rosenwald schools because they're named after Julius Rosenwald, who was a northern philanthropist. He was a really rich guy who- he was a part owner of Sears, Roebuck, the like department store. And he had a lot of different initiatives, sort of charitable initiatives in the South particularly, where his fund would sort of try to take on big social issues of the day. And so he, along with the Tuskegee Institute, sort of thought, you know, one big issue in this period is that there's just really low provision of education for Black students. And particularly, the South is very rural at this time. And so if there's a school, it's really far away and it's too hard to get to. And so there's a big gap in education between Black and white students in this area. And so they said, you know, we're going to try to build- or incentivize and provide some grants to provide schooling for students that don't have access to school. And so they they build about 5,000 schools through a matching grant program, where basically the local areas could apply for money and then they'd have to come up with some of it and they would give them a lot of the grant. So there were these really great schools where they were not only providing access to school, but they actually- they had their own way of building them. And so they had some sort of like higher school quality as well. They were good schools at the time. And they sort of really increased provision of education for these kids that didn't have any access to school.

 

Jennifer [00:03:46] So what was the rollout of these schools like across places in time? Could anyone simply request one for their town?

 

Katherine [00:03:51] Basically, so, I mean, if you look at the funding, it was a matching grant program. So you could you know, if you could come up with, I think, 60 or 70 percent of the money in the local area—so either the school district could come up with it or private organizations could also come up with money—and then once you did that, you could apply and you would get the matching funds. Then once you had the school, you were basically responsible for keeping it running. And so it's almost an incentive to build a school with a bit of money, but it's not like it covered everything. You know, people sort of looked at in various papers thinking about what areas actually built these schools. But there's you know if there's 5,000, they roll out over, I think 1915 or '16 all the way to about '32. And so there's a lot of variation across the country and sort of, you know, when they pop up in different areas.

 

Jennifer [00:04:32] And there there has been some research in the past about the effect of Rosenwald schools on educational outcomes. So what did that earlier work find?

 

Katherine [00:04:40] Yeah, so there's a big paper by Dan Aaronson and Bhash Mazumder, and they were the first ones to really collect these cards. Basically, like every school had its card in an archive and they collected those and got the year it opened, how many seats it had, and they sort of look at just the effect on years of schooling. And they show that basically there's unsurprisingly a big impact on completed years of school and also on literacy, they look at as well. You see big increases in enrollment and then you see effects on completed education later. So they said that essentially you think about sort of the gaps between completed education for Black and white kids, the Rosenwald school sort of close that gap by about half for about a year in this period. So that they basically contributed about a year of education on average to Black students in the South.

 

Jennifer [00:05:23] But your focused here on the effects of criminal justice involvement, specifically incarceration rates for Black Americans. So give us a little bit of context. What do Black and white incarceration rates look like during the early 20th century?

 

Katherine [00:05:36] Yeah, so it's kind of depressing in a sense, and not surprising either, that if you're Black in this early part of the 20th century you're a lot more likely to be incarcerated than if you're white. I think the ratio is about four times more likely to be in jail if you're Black than if you're white. And that's sort of similar to what it is today. It's changed a little bit over time, on and off. But the rate of incarceration's quite a bit lower then. We all know that there's been this big increase in incarceration in the US since the 80s. So in the early 20th century, rates were definitely lower. But there are still these huge racial gaps in terms of incarceration rates and conviction rates as well.

 

Jennifer [00:06:08] So what's the mechanism you have in mind for why these schools might affect incarceration?

 

Katherine [00:06:12] Well, I mean, it's always a complicated story, right? So there's a million different things out there in the crime education literature about why we might think that you can go from a kid going to school for longer to an adult not being incarcerated. The main mechanism that we sort of talk about sort of from the economic perspective is really just an opportunity cost story. That if you have more schooling, you're going to have higher wages. And if you're incarcerated, then you're giving up those wages and so well also hopefully you're not going to commit the crime as often. But I mean, that makes a lot of assumptions about sort of the other parts in which you can sort of end up incarcerated. And in particular in this period, very racist Jim Crow era, I think there's probably other things that are going on that I wasn't really able to speak to too much in the paper. For example, just being Black and walking down the street could get you arrested in this period. And so, like, there are these vagrancy laws. And so if schooling just increases the probability of being at work or having a job, that would also possibly just lead to lower incarceration. Right. You might just be better at interacting with the criminal justice system. You're in a better paid job, and so they're just- they're not going to lock you up for nothing.

 

Katherine [00:07:14] And there's a lot of different things that could be going on. It was really hard in this paper to get at much of that, just because I don't have- I see the outcome of incarceration. I don't see things like arrests and conviction. Right. There's other steps to getting- to ending up in prison. But it could be that you're just better at not being convicted or you've got some other way of not ending up incarcerated. I was trying to be very careful in the paper not to say "I'm not measuring committing a crime, I'm measuring ending up in jail." Or prison, which are two different things, particularly given the different mechanisms that I could work through.

 

Jennifer [00:07:43] Yeah, and incarceration is still certainly a very costly outcome for all involved. So important, important, even though it might not be a direct measure of crime. OK, so this question is obviously difficult enough to answer with current data in the present day. But you're looking at a period one hundred years ago. So what are the hurdles that you had to overcome in order to measure the causal effects of an intervention like this?

 

Katherine [00:08:07] Yeah, well, there's a reason this took eight years to write. Well, it took about five years to write then four years to get published. Yeah, so I mean, I was lucky in the sense that the Rosenwald schools were a nice identification strategy that I could sort of take off the shelf because there was this other paper and people had basically said this seems to work, comparing different groups and using what we call like a triple difference identification strategy. The biggest issue is really data. So in this period, in particular, figuring out how to measure crime or incarceration or anything at the individual level is almost impossible. And so that's where I spent a lot of time using full count census data before that. And so I knew that I figured out finally that they actually innumerate prisoners in the census and are able to to get them out of the census in the three census years that I use. You would want to just say what's the effect of education on incarceration? And that's a bit more difficult, obviously, because there's a lot of issues with why are people more educated than others? So using the Rosenwald schools as a way to sort of get around that and just look at this one program and sort of how that affects people's outcomes later on as adults.

 

Jennifer [00:09:08] Right. And so what you're kind of alluding to here is that, you know, if you wanted to know what the impact of these schools is, you couldn't just compare places with the schools, with those without. And so because those places might be different in other ways that are tough, maybe because you're motivated to to help the kids or reduce incarceration rates that you build one of these schools in the first place. So, as you mentioned, you're going to use this strategy that had been used before. You're going to use the gradual rollout of the Rosenwald schools across rural areas as a natural experiment. So walk us through that approach. How do you use the construction of these schools to isolate the causal effect of access to education on the Black children who lived in those areas?

 

Katherine [00:09:47] Yeah, so you're exactly right. So the first comparison would be probably not so great, which would be comparing places that have schools to places that don't. That's what we're ultimately trying to sort of get at. But we're going to difference out some other things just to sort of say let's let's focus on the group that was really affected and allow the other groups to sort of act like a placebo or control group for them. So here, your first sort of difference if you want, is students that are of the right age when the school comes versus students that are a little bit too old. And so you can sort of- we have like a measure of exposure to the schools based on how many years when you were 7 to 13, was there a school in your county. So like there's some students who that's going to be a zero for, because they're too old, and some students have variation there. So that's kind of like the variation in cohort, not just across different places.

 

Katherine [00:10:34] And then we also say, OK, it's only rural areas that get it. And so whatever is going on in the county as a whole that could be affecting rural and urban places, right, these are just richer counties or they're more likely to be growing cotton, anything like that. We're going to compare rural to urban areas and let the urban areas sort of wash out things that are happening at the county level. And then the third is that these are schools targeted to Black students. And so we use white students as an initial counterfactual group, where like if there's anything affecting children who live in rural areas, but all children who live in rural areas, then that should get washed out or taken care of by using the white children as your comparison. So here it's like we call it triple difference, where you're looking at different ages, Black versus white and rural versus urban. And you're saying only the students who are the right age at the time of the school are affected, only the people living in rural areas and only those that are Black relative to white.

 

Jennifer [00:11:23] Excellent. OK, so in order to do this, you need to know where people grew up as children and then you need to know if they were incarcerated as adults. And again, this would be difficult enough with the current data. I mean, honestly, I can't imagine even writing this paper about present day people. But so so tell us about the data you used for this project. How do you construct the dataset that you need?

 

Katherine [00:11:45] Oh I'm just laughing because it was so much harder than it would be now. So, I mean, let's just talk about census data in the United States for a second, and then I'll talk about specifically how I used it. So, now if you're a grad student or even an undergraduate, you can go to IPUMS and you can download the full count US Census data and it says, "Is this person incarcerated?" And that's part of what you need. You can also go and get the children and see where they live. For me, I had to really collect all of this incarceration data by hand. So I flipped through three entire censuses basically to code up who was in prison. So that was three years of work. But basically in the census, you can identify prisoners or you can identify people in there in the state or the federal prisons or even a local jail. And then the census data, which is great, as children 10 or 20 years before you can see the town they live in. And then the nice thing about census data in the past, and what sort of worked on a lot of papers, is that because it's old data, you've got names and you've got other identifying information like age and race and birth place. And so you can link people across censuses so you can see someone as a five year old in 1910, and then see them, as a twenty five year old in 1930. And as long as they've sort of not changed their name too much and they've reported their age correctly and they've kept the same birth place and they're quite unique—they don't have the same name and age and birthplace as someone else—then you can link them across those censuses. And so then you get their childhood location and their adult outcome, either in prison or not in prison as well as like where they live. I mean, there's obviously lots of challenges with linking.

 

Jennifer [00:13:14] Those ifs that you mentioned seems to be a lot of work in the census.

 

Katherine [00:13:19] Yeah exactly, and particularly so it's only men because women change the name when they get married. For Black men, it's a lot harder because literacy rates tend to be lower. And so they might misspell their name more often, they are less likely to report the right age. And so for Black men, it's about- the match rate is about half that it is for white men. But it is possible. It's just- yeah, this is why we need the full count census data. You've got one hundred percent of people and then you lose up to 90 percent of them in this process. But you're still left with hundreds of thousands because you've got everyone.

 

Jennifer [00:13:47] And say a little bit more about the process of- you said you could go and actually see them enumerated in the jails or prisons. What does that look like?

 

Katherine [00:13:56] Yeah so, I mean, in this period, it's not like today where you would like fill out you're own little census form. Enumerators would go around and they they go door to door to houses and then they also go to any institution. And so on the page, it'll say, like at the top of the page, this is the North Carolina State Prison. And then it will just have inmate or prisoner just written and they have their name, their age, their- all the information about them. And so you can just- like I didn't flip through the entire census. I had to basically identify pages based on like-relationships that were- that had been typed in the index, but which were like missing or prisoner. And I'd have to go through and just check that these people were actually incarcerated. But it's yeah, it's all handwritten stuff that the microdata would would have like, "Are you incarcerated?" And then you'd have all of the comparison group, obviously that's not incarcerated.

 

Jennifer [00:14:39] And you said today all of that is is just sort of incorporated in the regular data. How did that happen?

 

Katherine [00:14:47] Well, I like to argue that it's not quite as good as what I did.

 

Jennifer [00:14:51] I'm sure.

 

Katherine [00:14:52] Yeah, so Ancestry.com and Familysearch.org are like the two big genealogical websites that always had the census data as soon as it was available. So like for me, a big challenge actually was at the 1940 census only came out in 2012 because it only comes out 72 years after it's taken. And so I was literally sitting around going in the job market, going, "Can you guys digitize this quickly?" It was really stressful. It was like September and I'm like, give me the data. But basically, what these genealogical websites do is they want you to be able to find your ancestors. And so they went in to all the older censuses and they would digitize just the name, the age, the birth place, and usually the relationship to household head. And so you could search on that. And so that was sort of available. Well, what they didn't feel like doing was entering things like education or literacy or income or occupation, because that's not very useful for finding your ancestor.

 

Katherine [00:15:42] And so now that's the thing, IPUMS worked with Ancestry to basically—I'm not exactly sure who paid for this—but essentially they went in and they digitized the rest of the field. So now we've got- you can go download OCC1950 as the occupation code. That's available for everyone now, whereas in the past we'd have to go in and take a sample and just write down the course of occupations or anything we wanted to get out of that. So it's been- that's why I'm just laughing, because it's like the cost of using IPUMS- of using census data has gone way down in the last like five years or so as they've digitized all these censuses. So I'll still say, I think my data is a little bit better, that they weren't super great at digitizing institution information on the top of the page. And so I think the group quarters variable, which measures incarceration is actually slightly to a lot undercounting incarceration. When my data- I had a bit of a higher incarceration rate than you would get in the census, so. I told IPUMS about that, they have no interest in fixing it. So maybe someday I'll go in and do the rest of the censuses.

 

Jennifer [00:16:41] Just by hand.

 

Katherine [00:16:41] Yeah, you'd have to go by hand and say, are you catching these big prisons? But I think what I did show in my paper is that it doesn't matter if I- so like I did this paper and then like by the time it's actually R and R in 2018 or whatever, IPUMS data was available. And so I showed, you know, here's my two years of work data and here's IPUMS and their group quarters. And even though their incarceration rate's lower, I can still use their definition and it's fine, so.

 

Jennifer [00:17:03] For the effect of the policy. That's great.

 

Katherine [00:17:08] Yeah, the same effect. So I was like great, if only I'd just waited five years.

 

Jennifer [00:17:11] If you'd waited five years, someone else might have gotten to it first. So it's good that you did all the work. And we wouldn't have known if it mattered if you hadn't-.

 

Katherine [00:17:20] And I wouldn't have gotten a job, so.

 

Jennifer [00:17:21] Exactly. Exactly. All right. So let's talk about the results then. You have all this amazing data, you're ready to go. What do you find is the effect of being exposed to a Rosenwald school as a child on incarceration as an adult?

 

Katherine [00:17:35] Yeah, so thankfully, I found something. Although zeros are things that can happen. But the challenge in economic history is that you collect all this data, you spend years at it, and then it's just mush. But no, I do find that exposure to one of these schools reduces incarceration as an adult. I think it's- if you think about how big it is, I think full exposure reduces your probability of being incarcerated by about 1.9 percentage points. But if you look at OK, that's actually basically the incarceration rate at the time. But if you think about, OK, what's the average exposure of these cohorts. You think about sort of scaling that, it's a reduction in the incarceration rate of about 8.8 percent of the sort of mean. So these sort of reduce on the base of- reduce at about like less than 10 percent in terms of what's your sort of base chance of being incarcerated.

 

Jennifer [00:18:20] That's great. And so one possible reason that education might affect incarceration rates is that it might encourage people to move. So you consider effects on migration directly and what do you find?

 

Katherine [00:18:31] Yeah, it's interesting. I don't actually find much. I tried- I remember fighting with these results for a long time. You definitely might think that education would particularly contribute to the Great Migration in this period. So there's a big migration north, particularly of Black men, particularly my age range. So everyone's moving. But it doesn't seem, at least in my data, that extra exposure to a Rosenwald school actually led to you moving more often. I do see a positive number, it's just not statistically significant. And so it could be there, it could be that I just don't have the power to pick it up. Some other literature has said there's effects on migration, but they're using a different set of data, different years. So, I mean, it could just be that we're doing a different regression. But yeah, I was a bit surprised that I didn't see that because I thought that you would see that education causes migration. And in that sense, in this situation, that's probably not a good outcome for respective incarceration because incarceration rates are so much higher in the North. So that would actually sort of- education makes you move and less likely to be incarcerated. But if it puts you in a place with higher incarceration rates, that would sort of temper the effect. So I don't- yeah, I don't see anything significant, which was surprising to me.

 

Jennifer [00:19:34] Yeah. I'll also flag- so Ellora Derenoncourt was on the show a little while back. So I'll put a link to her her episode about the Great Migration in the show notes. So you also look directly at the other potential mechanism, education level, as proxied by literacy rates and years of completed education. So what effect did exposure to Rosenwald schools have on those outcomes?

 

Katherine [00:19:54] Yeah, so thankfully I was able to, in this situation, replicate what the other paper had found as well. They weren't using linked data, so I was able to do a little bit more detailed, really actual exposure as a child and actual outcomes as an adult. So I do see that full exposure increases education levels and that's only in the 1940 census and so you lose a little bit- so I'm using the 1920, '30, and '40 censuses as my outcome years. We don't collect education until 1940. And so if you look at just the 1940 census, full exposure increases your years of education by a little over a year. So you're exposed for seven whole years as a child, that translates to about an extra year of education. And I see similar things for literacy, which is available in all the censuses. It's available in '20 and '30, and then you can sort of guess at it in '40. And you see basically if you've gone to school for about two or three years, you're literate. And that's kind of how you back that up. But I also see that yes, literacy goes up by about six percentage points for full exposure. And this is on a base of- let's see if I have sample means. I do actually, it's on a base of 78 percent, looking at my table. So it goes up a decent amount for Black kids that are exposed, yeah.

 

Jennifer [00:21:01] And then you run a bunch of additional checks, primarily focused on the construction of your dataset since that was such an important part of this project. So tell us about some of the checks that you think were most important.

 

Katherine [00:21:14] Yeah, I think any paper that uses linked data, now in particular, has to basically run a bunch of different linking methods and just make sure that you get something similar. When you're linking, there's sort of a- there's a tradeoff between getting a big sample size, but possibly getting a lot of false positives or like you're you're going to call this person a match, even though they're not. And that can tend to attenuate what you find, depending on what your regression is. And so one of the robustness things is just to make yourself be more strict and say, OK, I'm going to make sure that you're very you're extra extra unique. There's no one near you in that age range. Your name's perfect, rather than sort of almost perfect. And that just sort of allows you to reduce more false positives and that you sort of see that that you get the same kind of thing. And, you know, basically what we've found throughout a lot of different papers is that the matching, it's always hold up. It doesn't really matter too much what you get, what you use.

 

Katherine [00:22:06] The thing that I liked that I was pushed by a referee was the sort of permutation test that they had me do, which I had to learn to figure out how to do. But basically, the idea is that you're worried a little bit that the timing of these schools and when they are introduced could be correlated with stuff going on in the county that you're not fully accounting for with your comparison groups. And that everything's just spurious. Like education levels are going up and incarceration rates are going down, it has nothing to do with a Rosenwald's school introduction. And so the way that referee suggested and we ended up doing is you kind of randomly reassign the schools at different times and different places. And you do this a thousand different times and you run the regression each time and then you sort of plot out those coefficients and you show that your coefficient is bigger in magnitude than almost all of those. And that kind of just- it's like no matter how you shuffle these schools around, you're getting the true effect when you do it and you're not getting like- the random effects that you get aren't nearly as big. Like, to me, that was like the most reassuring thing, that there is probably a true effect there. It's not just some spurious thing that I found.

 

Jennifer [00:23:04] That's great. And OK, so that's your paper. Have any other papers related to this topic come out since you first started working on the study many years ago?

 

Katherine [00:23:13] Many years ago when I was a young little-.

 

Jennifer [00:23:18] Economist.

 

Katherine [00:23:19] Yeah. Yeah, so I guess there's two literatures here. There hasn't been anything linking the Rosenwald schools and the crime in particular, incarceration in particular. I think that there's a few papers on Rosenwald schools. There's some thinking about sort of- well there's one by Carruthers and Wanamaker thinking about the funding and how essentially sort of the political economy of this money flowing to these counties is that they didn't want to spend money on Black schools. That was sort of the goal of a white school board. And so you actually- they see that essentially a lot of this money is sort of siphoned away. Right. An extra dollar actually means an extra dollar for white schools, not for Black schools, which is interesting. And so they have to sort of think about why you still see effects in education. I really like that because it thinks a lot about sort of the actual mechanisms going on politically in the local areas.

 

Katherine [00:24:06] In respect to the incarceration literature, I think there's sort of a thin literature in the early 20th century in terms of incarceration. I read a follow up paper that just looked at the Great Migration and Incarceration. Basically, if you compare brothers, one who stayed in the South and one who moves north, the one who moves north is more likely to end up incarcerated. And that's sort of because incarceration rates are a lot higher in the North, even though these two people sort of come from the same background, have very similar characteristics. So that was a really easy paper to write. I was like, I took my data and said you know, I can do a different- I can get a second paper out of this.

 

Jennifer [00:24:37] Yeah, I guess it's not shocking that there that many sibling pairs where one stays and one goes, but are there are a lot? Obviously enough to write a paper with.

 

Katherine [00:24:47] Yeah. So it's not a ton. In particular, because incarcerations are relatively rare events. I mean I was a little bit surprised I had an update. I think I had about twenty thousand sibling pairs that one stayed in the South and one moved, and a bunch that both stayed or both moved. I could look at like movement to the North and also movement to cities in the South and no matter where you move, urbanization is just- urban places have higher incarceration rates. That's just the way that that works.

 

Katherine [00:25:13] Yeah so there's that and then there's not a whole lot on this period—that I know of, at least—there's some work, I think, by some graduate students who are in political science or sociology looking at sort of convict labor. And this like- a little bit earlier, where before they start building state prisons, most people who are convicted of a crime would end up being rented out as laborer to a farm or a road building or something like that. And that's how states would essentially incarcerate without incarcerating. Like, they didn't have to pay for it to keep people in jail. They would just use them as free labor. So there's been a lot more- a lot of interest in that. And then also in sort of when they get state prisons, now you're incarcerating people in the prison, but it's also a productive prison. So like Parchman farm, one of the best examples of a huge state prison and basically, like they profit off of incarcerated people, like they just make you work and they don't pay you. So that's I think that's a big thing that people have been looking at.

 

Katherine [00:26:09] There's a paper I thought about writing about five years ago, and I didn't. I actually just talk to these people, I think in chronological time it was last week. But so you recorded a while ago, it's the new paper in the AER about the effect of your parents being incarcerated on children. And I remember seeing a draft of that maybe a few years ago. And I was like, OK, they're doing it better. I don't think I even read it at the time. But I was like, oh, they can do that really nicely. But I do want to do that in the historical context. I think it would be a really interesting contrast. I've got some really cool data from South Carolina, where they publish every single person who's accused of a crime and they've got the court date and they've got like their name, what they're accused of, whether or not they're convicted, their age, their name. Right. So things you could actually use to link them to the census and see if they have children and then sort of link those children and see what happens. So that's kind of, I think something I'd like to do. I sort of was like, oh, they're doing it in a modern context. They're going to do it better. But actually, I think more information there would be useful. So-.

 

Jennifer [00:27:05] Absolutely.

 

Katherine [00:27:07] -maybe I'll resurrect that idea.

 

Jennifer [00:27:07] Yeah, no, absolutely. I think both because the results of that paper are so striking. So this is- so Jeff Weaver was on the show early on, early in the show's existence—so I'll also put a link to that in the show notes. But yeah, they're basically finding that parental incarceration for those on the margin of incarceration seems to have at least some benefits for the kids, which is counter the, you know, the prior I think most people had and they had going in. And so certainly there's there's surely lots of heterogeneity across families and across contexts. But it would also be really interesting to know if this effect has changed over time.

 

Katherine [00:27:40] Yeah. I mean, my prior would be also that in the past that I mean, especially in 1920- I mean, you're taking a big income earner away from the household. So I think that income effect would really matter a lot.

 

Jennifer [00:27:51] Yep. Yeah I vote I vote that you write that paper.

 

Katherine [00:27:55] Alright, I'll see what grad student needs an idea to work on. Got any grad students?

 

Jennifer [00:28:01] Perfect.

 

Katherine [00:28:02] Man, that would be great.

 

Jennifer [00:28:03] Cool. OK, so what are the policy implications of your results and the other work in this area? What should policymakers who are listening take away from all this?

 

Katherine [00:28:10] Oh, I'm always very hesitant to say anything about the past and what you should think about today in terms of policy. I think it says that education matters. Right. I think we know that. We know that there's social returns to education, and that's it's definitely something that we should really think about when we're designing programs. I think particularly if you're thinking about how to target a social program. Right. Here we're targeting it to a group that's both at high risk of incarceration and also has very low access to schooling, you might get a really big bang for your buck. So I think it says something about targeting, but also just, you know, we know there's returns to school and- that are not just monetary. So I think just continue to add to that evidence is important to think about. Like, why should we even build more schools or give more- I mean, in this period, like everyone has a school, like why should we make schools better? I think there's a lot of other payoffs to that.

 

Jennifer [00:28:55] And what's the research frontier? We've already talked about this a little bit, I guess, but what are the next big questions that you and others will be thinking about going forward?

 

Katherine [00:29:03] Yeah, I mean, I haven't really been working on the on crime as much since this paper was finally finished, I was so tired of it. Yeah, I think the stuff we talked about, I think there's still questions that you might ask in a modern context that you could also ask in the past. And that—this is sort of the sell that I give to grad students as to why they should do history—is that sometimes you can use historical data and answer a question you might want to answer now, but that you just don't have the data for. And I also think US history is interesting. The other way I sell it to students is just saying that the US and particularly the South was a developing country in the 1900s. Right. It was quite poor, very agricultural. So if you're thinking of development questions and you don't necessarily have data, I think the US South is always a great laboratory to think about this stuff. So that's kind of my pitch for history in the why- like the big research topics are sort of, you know, anything you could think of you could probably use old data to do.

 

Jennifer [00:29:51] Yeah. And especially I mean, as we were just talking about, if you are interested in long run effects and especially intergenerational effects, like you have to go back a certain amount of time in order to be able to look at what happens to someone's kids or grandkids. And then the upside is you've got all the census data to work with, which sounds like kind of a dream, frankly.

 

Katherine [00:30:09] It's pretty rich, yeah. Yeah, I will also put a pitch there for history that I'm working on a totally unrelated project, but they've also now through the census restricted data centers, you can also link historical data to modern data so you can actually trace people in their intergenerational all the way from like 1910-now. They'll match the CBS in the 70s and like the 1990 census. And so, like, I think you could take some of these questions from the past and say, you know, what's the intergenerational effect of Rosenwald schools? I think someone's actually working on that, now that I think about it, using this data. That's the great thing about the past, you've got one hundred years in the future. So you can you can really think about like second generation, even third generation effects of stuff.

 

Jennifer [00:30:46] Yeah, well, and I guess if it's in the census RDC, does that mean you can link to the CJARS data that Mike Mueller-Smith is putting together? Like-.

 

Katherine [00:30:54] That's a very good idea. I don't know if that's- is that picked like the-?

 

Jennifer [00:30:57] Yeah.

 

Katherine [00:30:57] Oh, wow.

 

Jennifer [00:30:59] He's got the personal identifying- yeah sure! This is just going to be a dream once those data are- I mean, they're I think they're fully available now. It's just once people get access to the RDCs.

 

Katherine [00:31:07] Yeah. We've actually- we just got our project approved for looking at some other stuff. But nice, now we can look at criminal outcome.

 

Jennifer [00:31:15] Yeah. Get on it. This has been my main plug to grad students. It's like, look, it's going to take all of us faculty forever to get SSS clearance and all the rest to get at the RDCs. And so this is your advantage as grad students. You can get in on the ground floor with the CJARS data and write all the cool papers.

 

Katherine [00:31:33] Cool, well now my study seems ---. It only took three years.

 

Jennifer [00:31:38] Yeah, I know. Oh my God, I went through the same thing recently. ---. Yes. All right, well we can complain about SSS clearance all day, but I will leave it there. My guest today has been Katherine Ericksson from UC Davis. Katherine, thank you so much for talking with me.

 

Katherine [00:31:53] Thank you, it was fun.

 

Jennifer [00:32:00] 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. And thanks also to our Patreon subscribers. This show is listener supported, so if you enjoy the podcast, then please consider contributing via Patreon. You can find a link on our website. Our sound engineer is John Keur with production assistance from Haley Grieshaber. Our music is by Werner and our logo was designed by Carrie Throckmorton. Thanks for listening and I'll talk to you in two weeks.