Probable Causation

View Original

Episode 49: Michael Makowsky

Jennifer Doleac

Michael Makowsky

Michael Makowsky is an Associate Professor of Economics at Clemson University.

Date: April 13, 2021

Bonus segment on Professor Makowsky’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. Makowsky's work on how the minimum wage and EITC affect recidivism:

"The Minimum Wage, EITC, and Criminal Recidivism" by Amanda Agan and Michael Makowsky.

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 Michael Makowsky. Mike is an Associate Professor of Economics at Clemson University. Mike, welcome to the show. 

 

Michael [00:00:26] Thanks for having me. This is fun. 

 

Jennifer [00:00:27] Today, we're going to talk about your research on how policies that increase wages affect recidivism. But before we dive into that, could you tell us about your research expertize and how you became interested in this topic? 

 

Michael [00:00:39] Well, I have two lines of research, and the one that's relevant here is obviously going to be I do a lot of work in law enforcement and criminal justice and to some degree crime more broadly. I have a second line on social groups and how they cohere and how they recruit members and how they get them to commit to the group. And it's especially focused on religious groups. Hopefully, that isn't relevant to today's discussion. As for how I got involved in this topic, I think some of it comes back to how I tend to engage in research. So I like to start with what I think is an important question that maybe a lot of people are asking and then twist it, you know, 30 degrees just to see if there's an angle that's neglected, or maybe we just have this natural tendency to miss. So when economists and policymakers talk about the minimum wage, they very understandably obsess with the demand side. If the minimum wage goes up, will employers hire fewer people? And that's a really important question. That's like the core of economics. 

 

Michael [00:01:41] But comparatively, we spend so much less time thinking about the supply side. So when you stop to think about the decisions workers are making and whether or not the supply of their labor and entering the marketplace, there's this built in assumption that we don't even know we're making most of the time that we're only going to talk about "normal jobs," normal jobs in scare quotes there. We're only going to talk about the legal labor market, the market for criminal activity, the market for just fixing cars in your garage, the gray cash market or running a daycare out of your living room or spending your evening sliding a skateboard under an SUV and boosting the catalytic converter. The literature is fairly blind to those options, and those are all different ways you can earn an income and spend your labor hours. So my coauthor on this project is Amanda Agan, and she's a top shelf crime economist. She spends a lot of her time thinking about what does it mean to be in the labor market with a criminal record. And I think when she and I had our very first conversation about the minimum wage and crime, our discussion was framed in just a really atypical way. And that's what eventually led to this project with the minimum wage and the EITC and criminal recidivism. 

 

Jennifer [00:02:57] So your paper is titled "The Minimum Wage, EITC, and Criminal Recidivism." It is coauthored with Amanda Agan. And in this paper, you focus on two policies that aim to increase wages for low wage workers. So let's talk about each of those in turn. First, what is the minimum wage and what policy changes are you focused on in this paper? 

 

Michael [00:03:17] So the minimum wage is obviously a floor in how much an employee can legally be paid per hour. They show up at the federal, state and also sub-state level, you'll see county and city minimum wages. And we're interested in establishing a causal relationship between the minimum wage and whether or not it increases or decreases criminal recidivism. Now we are constrained by the data, and we'll talk about that in a little bit. But using prison data, we have to focus on state and federal minimum wage changes. We can dabble a little bit in the sub state minimum wages, things at the county and city level, but we can't really put our weight behind that because the data is just geographically limited. So the question is when a person is released from prison, does their probability of returning to prison depend on the minimum wage of the labor market they are released into? So we have data from 42 states between 2000 and 2014, and that gives us a range of minimum wages between $5.15 an hour, all the way up to $9.50 an hour. And in this window, we're fortunate enough that the average state changes their minimum wage or has their minimum wage effectively changed from outside four times. So that gives us a fair amount of variation. And I would also like to note that it's fun knowing that I have been paid a wage less than the minimum minimum wage in our entire dataset. 

 

Jennifer [00:04:42] Fun fact about Mike. Okay, and then next, what is the EITC and what policy changes are you interested in there? 

 

Michael [00:04:51] Now, the EITC is a little more complicated. It's the earned income tax credit, which is arguably the closest the U.S. comes to a pure government wage subsidy, at least a broad wage subsidy. So for low to moderate incomes, the EITC can refund a portion of their taxes back, and for a lot of tax filers, this can actually result in a negative marginal tax rate, meaning that their tax refund at the end of the year is greater than what was taken from all of their paychecks throughout the year. So it's scales, of course, with the number of dependent children you have, and this is really important. So if you have two kids, you can receive a federal refund of up to $6,000 a year. If you don't have any kids, though, that's capped at $500. So that relationship, it's going to be more complex when we try to interpret and make predictions about how this is going to affect decision making. 

 

Michael [00:05:49] Now, economists have an affinity for wage subsidies because they both put cash in the hands of low income households, but they also induce employers to increase their hiring. The catch, of course, is that politics rarely creates chalkboard economic policies and the EITC because it varies with the number of dependent children because there's this big delay between your earnings date and when you actually receive the refund and the fact that it depends on you being able to correctly file your taxes, this is going to put some hiccups in the process and make it a little murkier in terms of what kind of effect we should expect. So for the purposes of this paper and for just our own curiosity as policy motivated individuals, the EITC serves as an interesting but imperfectly parallel policy to the minimum wage. 

 

Michael [00:06:39] On the one hand, it's nice that the employment and the wage effects, they move in the same direction. So with the minimum wage, if you do an Econ 101 theory, you're worried that while it's going to raise some wages, it's going to remove some job opportunities. It's going to lower employment, potentially. You don't have to deal with that with the EITC. It should, in theory, all push in the same beneficial direction. And so we hope that that effect will lead to it reducing recidivism. On the other hand, because they depend on child custody, and that is something that is not guaranteed for released prisoners, even if they have children, particularly for men, it gives us pause in how confident we should be in the existing policy's effects. 

 

Jennifer [00:07:20] So just to clarify a little bit more. So with the EITC, basically, if you earn under a certain cap for the year and depending on these various factors like do you have dependent kids and how many, then for like every dollar you earn, you get an extra 10 cents from the government or something like that? 

 

Michael [00:07:36] Yep, that's exactly right. Yeah. So it's going to predominantly affect low and moderate incomes. It's going to start to fade out earlier if you don't have children. I think for people who have two children it's going to start to fade out around 50,000 pretty quickly. Now, for the purposes of our study, we're not going to use the federal EITC. We're going to use the 22 states that offer an additional what we call a top up or what the literature calls a top up to the federal EITC. So states have chosen - 22 states have chosen to add an additional 3.5 to 40 percent of the federal amount to the recipients' incomes at the end of the year. And this is - we can - when we talk about the identification strategy, we have six states, three of which added in EITC top up in our data window, that's North Carolina, Nebraska and Oklahoma. We also have three states that unfunded their program in Colorado, Indiana and Michigan. So we're able to use those six states and say what happens to recidivism within the states that add a program versus the ones that remove theirs? 

 

Jennifer [00:08:40] Great. And just to elaborate a little bit more also on why economists like the EITC. I mean, there's something I think economists are really hesitant to mess with prices as a signal. Right. So there's this idea that like a market wage is the combination of supply and demand and depends on your skills and productivity and all that stuff. And if we as a society think that someone should be earning more than the market wage, then we can, through the tax system, decide that if you're earning $10 an hour, we can make sure that's actually $15 is what you take home. And of course, but then it's complicated for all the reasons - the implementation is tricky. 

 

Michael [00:09:20] And it's nice too, because I mean, all policies have their strengths and their shortcomings. So the EITC has the benefit of all the incentives are moving in the same direction. Nobody has an incentive to non-comply. But the minimum wage has the benefit of it's just so simple. This is the least you're allowed to pay someone. You don't need help with your taxes. You don't have to do any math. It's just so simple. And also, you know, from a political point of view, it's an easy thing to sell to voters. The EITC is always going to be a little trickier, which means that I mean, both programs are always going to be caught up in the political climate, which is going to be part of why it's hard to identify these effects, part of the reason we have to be so careful when we do this statistical analysis. I mean, to give you a sense, I'll give you an anecdote a friend of mine gave me who worked on the Hill for years. He said the EITC is an incredibly successful program and you can tell that because nobody ever talks about it. And this is a different political climate. This is in the 90s and early aughts, but there was an understood truce that we'll never talk about the EITC as Democrats or Republicans because it's working. And if we talk about it, we have to disagree about it. 

 

Jennifer [00:10:35] That said, I think there is interest in expanding the EITC, which then makes your study quite relevant because we want to know if - how it's working and how it's working for different groups and so we can get into all of that later. But in the meantime, let's talk more about your paper. So why are these policies of interest to folks who care about criminal recidivism? What are the mechanisms that you and Amanda have in mind for how they might change behavior? 

 

Michael [00:10:58] So this is where it starts to get really fun for me. Let's leave that econ 101 framework behind for a second. And yes, we know that a minimum wage should raise wages, but it could decrease employment depending on the - what we call the elasticities, the sensitivities of employers and how much they might reduce employment, or also for people who care about the criminal justice system, the minimum wage is going to create, in theory, a surplus of workers applying for jobs, which gives employers a surplus of applicants to choose amongst. So if they have a longer line of people applying than people they need to hire, or they anticipate hiring, they can use their discretion. And one group of people were always worried are going to be sort of at the low end of the totem pole in that discussion is people with a criminal record. There's just a mountain of evidence that the stigma of a criminal record in the hiring process is significant. 

 

Michael [00:11:52] But when we focus on recidivism, so people who have a criminal record and have now been released from prison. So everybody in that group has a criminal record. They're all facing the stigma. We know that they may or may not command a lower wage in the market, but we also need to remember that they no longer have to worry about the opportunity cost of participating in the illegal labor market. They already have the Scarlet Letter. One more crime isn't going to matter that much at the margin if they get arrested, so they're going to look at their decision - even if we start in a world where let's say the market wage is 10 bucks an hour and then the minimum wage is 15. And that leads to less employment. But that person who's been released from prison, they look out at the world and go, well, the market's paying me dollars, but I can sell weed, I can boost palladium, I can boost, you know, sweaters at TJ Maxx, I can participate in illegal sex work on the side. I have a criminal wage I can earn of $12 an hour. That becomes my reservation wage, meaning that if the market can't beat $12 an hour, I'm not interested in making $10 an hour in the legal market when I can make 12 in the gray or illegal market. Well, in that world now the market pays me 10 crime pays me 12, if the minimum wage is 15, then any disemployment from that minimum wage is irrelevant to me because I was never interested in those $10 an hour jobs, but $15 an hour that's enough to pull me out of the illegal labor market. 

 

Michael [00:13:31] So the disemployment effects, while hypothetical, are irrelevant to me. The wage effect, though. Now that's real. So you can imagine a world now where a minimum wage, if it's greater than that criminal wage, then we can actually see recidivism rates fall as release prisoners participate less in illegal labor markets. Now we can then compare that to the much, much simpler story of a wage subsidy of the EITC and say, well, we don't have to tell that complicated story. We can just say there are no disemployment effect. There's a market wage. If the EITC is large enough, large enough that it pushes the market wage above the criminal wage, it'll pull me out. So there's no countervailing effect there. It just has to be big enough. 

 

Jennifer [00:14:19] Right. And so when you're saying the disemployment effect is irrelevant to the person deciding whether or not to look for a job. But of course, it's relevant about determining whether they can find a job at $15 an hour, right. So if enough employers say if I pay $15 an hour, I'm going to hire this college grad, not someone with a criminal record. Then they might want the job, but they're not going to get it. 

 

Michael [00:14:41] So you can think of then with the minimum wage is just now above the criminal wage and the market wage, it becomes a sort of lottery at that sort of low end. And then as it gets higher and higher, the other options of the employer become relevant. You know, I can replace you with capital equipment, I can replace you with skilled - higher skilled labor. So you can imagine that there's this countervailing wage and disemployment effect, and we know there's some minimum wage, you know, let's pick some round number $100 an hour at which the disemployment effect is going to be so enormous it doesn't matter how attractive the legal labor market is, a low education person with a criminal record is just not going to be competitive for that wage. Right. So we have these two countervailing forces. What's nice here is we've gone from an econ 101 framework which just says less employment and it's going to disfavor people with the criminal record. We've now expanded it, and now the effect becomes ambiguous and that there is some range of minimum wages that we should expect a net reduction in recidivism. And it's at least feasible in the range we're considering because in our range of data, we don't see any especially outsized increases in the minimum wage. We're not looking at 1938 Puerto Rico, where the minimum wage suddenly overnight became several hundred percent of the median wage in the island. So we're actually - by focusing between 2000 and 2014 on just these 42 states it's genuinely feasible. We could be within that sweet spot range where the minimum wage is going to reduce recidivism. 

 

Jennifer [00:16:18] And remind us what the cap was again. 

 

Michael [00:16:20] It's a 9.50. 

 

Jennifer [00:16:22] Okay. And the other factor here that - kind of the main story you're telling in this paper is kind of this push and pull of how many jobs are lost to this group versus how many additional people are pulled into this legal labor market. But it also - the people who still have jobs, even if they had jobs before and were always going to work in the legal labor market, some of them, maybe most of them are earning more money now. And that also could reduce criminal recidivism, right. I mean, I think a lot of our models sort of frame it as a choice between crime or illegal work. But of course, you can have legal work and still commit crime it's not necessarily an either or. 

 

Michael [00:16:58] I don't want to get into like the incapacitation effects of just being busy at work so much as you know, crime is, I suspect, for most people, a side hustle. I mean, don't get me wrong, if you're writing a show for FX, you should tell the story of crime as your principal occupation because it's far more interesting dramatically. But in real life, it's always behooves you to have some sort of connection to the legal labor market if possible, particularly if you're on parole. There's a certain obligation to always be actively seeking work if you don't have it. But whether or not you feel it's worth the risk to supplement your wages in this sort of side hustle crime market, I think the model ends up being the same. It's just from the point of view of explication it becomes a lot more complicated. 

 

Jennifer [00:17:43] Yeah. And then, of course, just to add one more mechanism to the mix here. You know, while we're talking about it, you know, not all crime is financially lucrative. And so if you now have more money and can go out to the bars more often and get drunk all the time, you could wind up in some bar fights and that could get you locked up. So people do a lot of dumb and violent stuff when they're drunk or high. So if you have more money, people do worry sometimes that giving people more discretionary income can lead to more abuse of substances. 

 

Michael [00:18:11] Well, yeah. Spoiler alert I can spoil the results a little bit here and say that we actually don't find any relationship with violent crime. And I think that's always - I'm reticent to mix our consumptive and our productive models. I'm always for expanding our models a little bit in terms of our human behavior. I mean, that's literally what we're trying to do in this paper to consider it more broadly. But things can get unwieldy, real fast if you try to tell a consumptive story along your productive story. That's more advice for grad students who are pursuing a dissertation, but it's something I always try to keep in mind. 

 

Jennifer [00:18:49] Fair enough. Yes. Of course, the real world doesn't always care about that. 

 

Michael [00:18:55] Not at all. Not even a little bit. 

 

Jennifer [00:18:59] Okay, so before this paper, what did we know about the effects of employment and higher wages on recidivism? 

 

Michael [00:19:06] So for me, there's two definitive papers, and I feel like I'm automatically being unfair saying that to what is a broad literature. But Crystal Yang has a 2017 paper called "Local Labor Markets and Criminal Recidivism." We were obviously inspired by the very direct title, and what she finds is that during the upswing of a business cycle, there's a net reduction in recidivism rates. So she finds a net reduction of about six percent in recidivism rates during the upswing of a business cycle. And she was able to, through her identification strategy, really pin that specifically unemployment and wage effects. And then there's a second paper the next year by Kevin Schnepel called good jobs and recidivism and what he looks at is he actually segments the market in terms of the types of jobs - the industry of jobs and what he finds is if construction and manufacturing have an uptick in jobs that for every increase in either - every new construction job or new manufacturing job per thousand people that you see a 1.8 and one percent respective reduction in recidivism rates within those labor markets. And I'm a big fan of those papers because they're just - they really hit the bullseye. They pick a specific thing they want to measure. And then they spend the entirety of the paper convincing you what you're observing is real and that the causal relationship is reliable and you walk - I mean, I walked away from those papers really convinced that our project was worth pursuing because they were able to convince me that when the labor market is more favorable, particularly to lower skill, less educated, males, that we saw a reduction in the return rate to prison. 

 

Jennifer [00:21:00] Yes, so if you happen to be released from prison at a time when the local low skilled labor market is really hot, when they're like lots of jobs available and lots of good jobs available, then you're less likely to go back to prison. And so this makes everyone say, well, how can we create that? Well, we don't have direct control over the local labor market. Unfortunately, in terms of how many jobs are available, we wish we could wave a magic wand, but we can't. So is there some way we can recreate that with policy. And so that leads to my next question about what we knew before your paper about the effects of the minimum wage and the EITC in particular, both in terms of recidivism, but also in terms of employment, since it's relevant for you? 

 

Michael [00:21:40] There are so many ways I could frame my answer that would fill the internet with rage directed at me and you. 

 

Jennifer [00:21:52] Let's not do that. 

 

Michael [00:21:53] Instead, what I want to do is I want to point you to two separate reviews of the literature that come from slightly different angles and draw slightly different conclusions. So Arin Dube in 2019, he has a - it's actually, I guess, a white paper he wrote up called "The Impact of Minimum Wages." And what he finds is if you look across the swath of literature that he finds reliable, the minimum wage is up to about 60 percent of the median wage in the relevant labor market don't have measurable significant disemployment effects. So that would equate to a national minimum wage in the U.S. of about $13 an hour. Now you've got to be careful because the U.S. is a very, very big country with a lot of labor markets in it. So the Arkansas labor market is very, very different than the Bay Area. But nonetheless, he presents pretty compelling literature review that says, look, there's a relative insensitivity on the demand side to the minimum wage within a reasonable range. 

 

Michael [00:23:02] And then David Neumark and his coauthor's last name is Shirley, they came out with a review of the literature about 20 minutes ago. It's actually like a month ago, I guess, where they find that a preponderance of the literature does estimate a negative effect. But they're careful to point out that the negative effect is really only of a magnitude that we're going to be interested in or find significant for teens and young adults and people with a relatively low amount of education, and that the evidence from studies where we look at workers for whom the minimum wage is going to be definitively salient, that we do see negative employment effects and that in these sort of low wage specific industries, the evidence is pretty one sided. 

 

Michael [00:23:49] So Dube says the minimum wage isn't having an effect up to a reasonable threshold and then Neumark and Shirley is saying it does have an effect. The only thing I would note is even on the Neumark side, they're not finding evidence of our historic minimum wages of the last 40 years outside of specific stories like 1938 Puerto Rico. They're not finding evidence of enormous negative employment effects. In fact, you could even make an argument that maybe we want to pull teens out of the labor market so that they will spend more time in school building human capital. But their literature, less than the one that Dube is presenting, does point towards there is a minimum wage on the horizon at which we could see particularly negative effects. 

 

Michael [00:24:37] My read of the literature, which is, you know, should be viewed as less reliable than Neumark or Dube because they spend their entire research lives mired in this utterly miserable debate. I really don't know how they do it. They have a greater emotional tolerance than I do. My read from it is the effects are relatively mild for what we've looked at. And unless you are a strong proponent of teen labor and its benefits to us growing as human beings that you shouldn't be too worried about the minimum wage as we have observed or currently observe. But you are well within your rights to get a little persnickety when somebody on Twitter suggests we should have a $30 an hour minimum wage. 

 

Jennifer [00:25:21] And there is not as much research as you, as you mentioned, like when you look at the folks who are working in these jobs, when you really focus on the marginal workers beyond teens, there's more evidence of this employment effect, I think, which is the relevant population for you. 

 

Michael [00:25:35] So much of this comes down to how you stratify your data and who you are particularly interested in identifying an effect for. 

 

Jennifer [00:25:43] Okay, so do we know anything about the minimum wage's effect on recidivism? 

 

Michael [00:25:47] At the time of us looking at that I didn't - I mean, I'm sure I'm going to annoy somebody - I didn't find anything in the literature that I found even remotely convincing, and it was always second or third order connected. 

 

[00:25:58] So I think there's a paper using analyst y data and looking at like if you say - if you reported having a gang affiliation, I think and then like, what's the effect of the minimum wage on you? I think they found an increase in recidivism. But there are so many caveats like that - I agree that there's - I think there's basically nothing using the kind of data that you have, but I do want to give those authors credit. 

 

Michael [00:26:20] Yeah, I wish I could remember their name. It's always tricky when someone is looking for something as I don't want to say as narrow, but as specific as gang affiliation. 

 

Jennifer [00:26:30] I hear you. I think it's sort of the limits to the analyst y. We'll put - we'll put links to all these studies on the show website. But yeah, tough to study and we'll get into why in a moment. So let's move on to the EITC. What do we know about the employment and recidivism effects there? 

 

Michael [00:26:45] There hasn't been a previous literature connecting it to recidivism, but people were worried that there might be disincentive effects or disemployment effects for males that women might outcompete them in the market because they're willing to work for a lower wage because they're more likely to have custody, that it could have effects on the severance of marriages. You could see how there would be a pathway to unexpected effects as related to crime, particularly if you're interested in things like household violence. If it's more likely to lead to dissolution of marriages or financial dependance within a household, and that can, you can see a pathway to that. 

 

Michael [00:27:28] Most of the literatures I've read haven't found any evidence to support those sort of concerns. The only thing I can say about the EITC is it seems to be working as intended. If anything, the dollar amounts are relatively mild, depending on what state you live in, and because it scales with the number of children, you know how many households this is salient to is going to be limited, especially when you start thinking about crime. 

 

Jennifer [00:27:55] Yeah. So this is mostly going to target women, basically. 

 

Michael [00:27:57] Yeah.

 

Jennifer [00:27:58] That's the relevant factor here that you're sort of dancing around. 

 

Michael [00:28:02] The hypothesis we went into going into our analysis - when we look at our data the first thing we have to acknowledge is we just have a lot less women in prison. So we're just not going to have the same precision necessarily broken up across 42 states, especially when only six of them - we're going to have what we call within state variation. So our minimum wage results will be identified off of over five million observations. But when we talk about women specifically now, we're down to about 800,000 observations. And now when we talked about the EITC and women, we're going to be down to about 80,000 observations. 

 

Jennifer [00:28:40] Yeah, what we need is an expanded EITC to adults without dependent children, which will include men. And then we can go test this. 

 

Michael [00:28:48] You - yes, that is - when I went through the literature and after we, you know, did our own analysis. If you said, Mike, what is the one policy you would advocate for wholeheartedly without any reservation, it is removing the dependent child condition of the EITC. 

 

Jennifer [00:29:09] Yeah.

 

Michael [00:29:09] That this is such an obviously effective policy that it helps people for whom the value of the marginal couple hundred dollars, let alone $1,000, is so high that the sensitivity of their household decisions is so high that if there's just there's no empirical or theoretical reason not to expand this program and to remove this sort of differential gender effect. 

 

Jennifer [00:29:36] Okay, let's come back to this at the end. Let's move along with your paper. So you've already alluded to this a bit, but what are the hurdles when you and Amanda started thinking about this topic and you're like, okay, how are we actually going to study this? What are the hurdles that you had to overcome in order to measure the causal effects of these two policies? 

 

Michael [00:29:53] Well, for both of these, at the end of the day, we need to try to distill out the effect of the minimum wage and the EITC from everything that's happening in parallel with them. So this is a particularly difficult question to answer if you're looking at what we would call a panel analysis. Maybe you're looking at the average recidivism rate within a county or a state that's going to just be too coarse to really learn a lot about these policies. So for the minimum wage, the principal obstacle is data. For the EITC, it's we have this additional challenge that we need policy variation. So you need to have not just a lot of detailed data, but you need to have it over a window of time where there was meaningful variation in the EITC. 

 

Jennifer [00:30:40] Right. And even in both cases, you do want to be careful like, you know, the naive comparison of just comparing people who live in a state with a $12 minimum wage versus people who live in a state with a $10 minimum wage and saying, oh, recidivism is lower in the first one, that's not necessarily going to tell you that it's the minimum wage that caused that there could be lots of other differences between the states. And so you do really need these like plausibly exogenous changes, right, or changes that you can argue or show in the data look random relative to the underlying trends in recidivism to convince yourselves that you're isolating the effect. 

 

Michael [00:31:15] Right. I mean, so again, I'm all about spoilers today. When we think about the EITC, we observe between state differences. So states with a higher EITC or with an EITC at all, women have a lower recidivism rate than states without one. But we can't really hang our hats on that because those states can be very different places. And it's entirely reasonable to believe that states that have a higher EITC may have a completely different criminal justice system and institutional treatment of women within that criminal justice system. So, yes, we observe these differences between the states, but we don't actually know that it's the EITC that's driving them, whereas for our analysis, we're going to focus on the within state effects. And the reason we want to do that is it's reasonable to believe that there's probably not going to be such enormous institutional and cultural shifts within a state before and after they enact an EITC, but that cultural or systemic shift is what's driving our results that as long as we do our best to control what's happening in the labor market, in the state and the differences within the individuals, we can be a lot more confident in that within the state effect when we observe it. 

 

Jennifer [00:32:30] Yeah. And one more point on the data front that I think often surprises people. There is no administrative or large scale survey data set that tells you whether someone has a criminal record and also if they are employed. 

 

Michael [00:32:45] Do you want to just jump right into the data and we talk about what we have? 

 

Jennifer [00:32:47] Let's do it. What data are you using? 

 

Michael [00:32:49] So we're using the National Corrections Reporting Program, the NCRP from 2000 to 2014, and it's all administrative records. So these come from the states. So the state prison systems then provide their administrative records to the Bureau of Justice Statistics, which then processes the data for extremely grateful researchers such as myself and Amanda. The good news is that gives us the day they're released from prison. We know the day they enter prison and we know the day they reenter prison. We know the crime they are admitted for. We know the court that process them. We know their education, their race, their gender. We know a lot about them. And you already hinted the downside here is we don't know their employment. So we're going to pursue a little side and a little bonus analysis at the end, which we - I'm sure we'll talk about to try to get at employment mechanisms. But for here, all we can treat this as is, we're observing people enter, leave and then potentially reenter the prison system within a state. So that's what we got to go on. 

 

Jennifer [00:33:59] Great. And you do have to make a few restrictions to this data set, including dropping California. So why do you do that? 

 

Michael [00:34:07] This is the part of the discussion we call graduate student advice corner. 

 

Jennifer [00:34:10] Yeah.

 

Michael [00:34:10] Dropping California from this data set is a story I tell - I think I'm confident I've told every single Ph.D. advisee I've ever had, which is why you need to learn about the institutions behind your data if you want to make a proper analysis. So in this case, for nearly half of our analysis window, California prisons were full and they started sending people to local jails instead. To the point of which, in 2011 they enacted a policy called the Public Safety Realignment Act that allowed people to serve their sentences in local jails. Well, our data's from prison administrative records. So if you entered or read more importantly, reentered after a re-arrest, if you enter jail and not prison, you're completely invisible to our data. So from the point of view of our data, essentially California at some point stopped putting people in prison. So I mean, that's the big one we have to do. We have to drop the entirety of the state of California. But we also have to drop observations if the record doesn't include your gender because we know that the EITC is going to so strongly correspond to gender in terms of its effect, potentially. If we don't know what crime you were sentenced for, if you weren't yet released from prison by the end of our window, we don't - maybe you were about to recidivate. We just don't know it. And if you died in prison, which you know to put up McCabe, take on this. Yeah. So if you're from California or we don't observe some really key aspect of the individual, we have to drop them from the data. 

 

Jennifer [00:35:45] Okay. And then what outcome measures are you focused on? 

 

Michael [00:35:48] We're principally focused on just whether or not you within our window reenter prison within one year of being released and three years of being released. So we include both results in the paper. We encourage readers to emphasize the three year rates, and the reason for that is you're going to observe more of their sort of long run participation in the labor market. You're going to be less sensitive to sort of, you know, short term hiccups or maybe just luck. But also, most importantly, there's such a lag on the EITC because it's a two policy paper and people naturally want to compare the effects of one to the other. You really can't evaluate the EITC at all with one year recidivism because you're not going to get that money probably till March February at the earliest if you're really proactive. And then there's going to be a subset of people for whom that's just not going to be salient. So if you get released in December of the previous year, you have absolutely no refund. You can expect that March. So yeah, we tend to emphasize the whether or not you reentered prison. So that's going to be our definition of recidivism, whether you reentered prison within three years of your release date. 

 

Jennifer [00:37:04] Great. And yeah, I mean, I think the other reason to focus on the long term outcome is the legal process can also take a while, right. So if you're rearrested in six months, you might not be incarcerated by the end of the year. It might take a long time. So - but yeah, I think the main reason I bring this up is just, you can't see if they're rearrested, you can't see if they're convicted and put on probation. You only see if they're back in prison. So that is - that's a reasonably high bar, but an important bar that we care about for policy reasons. Okay, so you are going to use the timing of when individuals are released from prison to determine whether they would be affected by a higher minimum wage or the EITC. So walk us through your approach. What's the ideal experiment that a researcher would run here to measure the causal effect of these policies? And how does your approach approximate that experiment? 

 

Michael [00:37:54] Hmm. Well, if I'm now the Czar of the U.S. and can create my own data environment using the levers of American policy, I feel very powerful. I enjoy this. So first of all, the analyst y has now expanded to everyone. So we're going to have longitudinal data on everyone. We're just going to track you as you specifically go through your labor and educational history. And then you're, you know, would be, you know, entering and exiting of the criminal justice system. But where I'm probably going to get a lot of pushback is I'm going to take a deck of playing cards and I'm going to write 10 percent, 25 percent and 50 percent on each card. And then I'm going to make the governors of all 50 states and the mayor of D.C. show up with me on television. They're each going to pick a card from the deck. And then that will establish their minimum wage as the percent of their state or city's median wage the previous year. 

 

Jennifer [00:38:49] I like that this is not only an experiment, but a game show. 

 

Michael [00:38:52] Oh yeah. I mean, mostly I'm here for for entertainment value. So now everyone has to commit to this for five years, and then I'm going to have a second deck of cards and I'm going to similarly write different percentages for the state EITC or city EITC top up. And so we're going to - they're going to pick a second card and now we're just going to randomize these policies and we're going to track people. And we're not just going to learn now whether or not it affects people's entering prison and then when they exit, will they reenter again later? But we're going to be able to track their labor decisions. We're going to able to track their educational decisions. We're going to also be able to geographically track whether or not does this push people with a criminal record towards different states? Do we see people with a criminal record now gravitate to the states with a high minimum wage? Because that's the only thing that's going to pull them out of the criminal market? Or do they really gravitate to the low minimum wage states where it's easier for them to compete for a job? And similarly, are we going to see people gravitate towards the EITC because they want that top up? Or do they - are they just completely ambivalent to it because they hate filing taxes? I don't know. It's - I mean, it'd be fun.

 

Jennifer [00:40:06] Yeah. It would you would be a lot of fun. Okay, so that's your ideal experiment. And how does your approach approximate that experiment? 

 

Michael [00:40:14] Well, we have to settle for the chaos of modern politics and policymaking, and we just have to take the variation we observe knowing full well that the different policies that we get are going to reflect the culture and the institutions and political vicissitudes that affect labor markets and crime in the criminal justice system and how they treat people in that system, how they treat the rehabilitation process, the options they see in the market. And we're going to accept the fact that we just can't know everything about those. All we can do is what we call it - we can control for it. So in this case, we're going to take the state as our minimum unit and we're going to say, well, how does the change in the minimum wage - there's a change in the EITC effect the re institutionalization of released prisoners within that state across our window.

 

Jennifer [00:41:09] It seemed like the trends were pretty flat through that policy change in terms of recidivism. There wasn't other stuff changing at the same time that would make us worried that that other thing is confounding your effect. Of course, if you actually compared people released from prison like one day before and one day after, they're both treated by the new minimum wage. So do you have some sort of - who exactly are you comparing here? 

 

Michael [00:41:32] I mean, we're essentially comparing people who were released within the state before the minimum wage change to people who were released from prison after the minimum wage change or after a minimum wage change. Or maybe a better way of putting it is were released under different minimum wage regimes. So people who were released in the $5 regime to the $7 regime, to the $9 regime, you know, because most of these states within our data are going to have four different minimum wage regimes. Right. And so we're going to compare across those, what's the trend? What's the relationship? How do people - comparable people with the same educational level, with the same gender, with the same race - I mean, released for the same crime, how did they compare released in regime A, B, C or D? 

 

Jennifer [00:42:21] Yep. Okay. And so you're kind of implicitly assuming here that like that initial wage when they're released is the most relevant, even if like six months later, it changes or something? 

 

Michael [00:42:31] Right. And so when we like a lot of labor economics papers these days, there's the main table, which we think is the best representation of the results. And then there's 75 pages of alternative specifications and robustness checks. And this isn't hyperbolic. This isn't unnecessary. There's just these are really complex environments. And as we just discussed with our idealized experiment that can never possibly happen, it would be grossly immoral in a lot of ways, that we have to make do with what we have. That means that we need to be extra careful to test all the possible alternative angles and really put the results through for lack of a better term as a stress test. Can it hold up to this sort of statistical abuse? And in our case, we're pretty convinced that it does. 

 

Jennifer [00:43:24] Yeah, I'm a big fan of the 75 page appendix. We'll get to that. All right. Let's talk about the results first, all in one place. So what do you find as the effect of increasing the minimum wage on the likelihood of re-offending? 

 

Michael [00:43:36] So what we observe is when the minimum wage is increased by a dollar that we see a 1.53 percentage point decrease in three year recidivism rates. And this effect is consistent across men and women. So we don't see any gender stratified differences. So let's put this in context. So the overall three year recidivism rate is about 35 percent. Okay, so about one in three people within three years if they're released from prison within three years, they're going to be back in prison. And your typical - your average minimum wage increase in our window is 50 cents. So the average minimum wage increase reduced recidivism by 2.2 percent, 2.3 percent. 

 

Jennifer [00:44:19] And so that's in line with the hypothesis that this like pulling some people who otherwise would have been in the illegal labor market, you're now pulling them into the legal labor market and that is outweighing any disemployment effect. 

 

Michael [00:44:33] Correct. That's correct. And I want to put that number in context. The 2.3 percent for your standard minimum wage increase, I'm sure to a lot of people that sounds almost trivial. But if you put it in the context of you go to the grocery store, there's 50 other people in the grocery store. One out of those 50 people isn't going to go to prison again because of this policy. And then you have to remember the weight of that consequence. 

 

Jennifer [00:45:02] Yeah, and reducing recidivism is hard, really hard. A lot of very well-meaning, very popular programs that do not work. So this is finding anything that reduces recidivism is important. 

 

Michael [00:45:15] Part of the reason we discussed the great Yang and Schnepel papers, which, by the way, should be a name of the band. But those papers, they're very convincing that the labor market matters for recidivism. But the thing I sort of skipped past is that we don't observe a lot of effect of policies literally designed to reduce recidivism. You have these really complex, really well-intended, genuinely noble policy constructs that involve a lot of time in human resources, and they are grudgingly acknowledged to largely be ineffective. So if you're throwing a lot of human time and energy at a problem and it's not working, and then you have a really simple labor policy that's taking one out of 50 fewer people going back to prison, that's actually a big deal. 

 

Jennifer [00:46:05] Yeah, okay. And then the second piece of your results, what do you find is the effect of increasing the EITC on the likelihood of re-offending? 

 

Michael [00:46:13] So not too much of a surprise for us. We don't observe anything for men. For women, we see states that introduce an EITC, that there's a two percentage point decrease in the three year recidivism rate. So it's really comparable to what we observed with the minimum wage. But again, we don't have the same depth of variation. So we don't, and again, we just have fewer women in the data. What we are confident saying is that it does have a beneficial effect for women. And that effect, given the state of the EITC, is much stronger for men. 

 

Jennifer [00:46:54] What types of reoffenses are driving these changes? 

 

Michael [00:46:57] So it's being driven by property and drug crimes for the minimum wage, and it's being driven by both men and women, as we said before, relatively even effects. And it's not surprising because those are for men and women income driving crimes. So property crimes, whether you're boosting - boosting meaning you're either shoplifting or stealing property for the intent of then reselling for profit or you're engaged in the illegal drug market. We see a lower rate of re-offense and reentering prison for those offenses. We don't see anything for violent crime, which I think again speaks to us as this is a policy where the mechanism, whether you know, there's wage and employment effects, it's channeling through the labor market and labor market decisions. Now when we look at the EITC effects for women, we still see the drug crime effects, but the effect remains negative for property crime, but it's a lot less precise, which is to be expected just because we have a lot less women. What we see show up, though, what unfortunately with our data has to get bundled under other which is non-drug property or violent crime. And it's this is speculation, but I think it's reasonable to speculate that because we're - this appears to be income driven effects that this is probably women engaging in sex work to generate additional income and the EITC is reducing their participation in that income mechanism. 

 

Jennifer [00:48:24] Okay, let's talk about that appendix. You run a bunch of additional checks to convince yourselves that your empirical strategy is isolating the causal effects of these policies. So tell us about some of your favorite checks and why you think they're helpful. 

 

Michael [00:48:40] So what I think we should do is separately record this portion as a forty five minute appendix to this podcast, and then we'll just talk about this, or we could do it in brief. So, all right, let's just go through a couple of them. So we include what we call polynomial time trends, where we throw on the right hand side essentially linear, quadratic and cubic and fourth order polynomial functions that are essentially trying to control for the business cycle, but a state specific business cycle. So we pulled this from Neumark 2014. He strongly advocates for this as a right hand side control because there's a lot happening in the labor market that could be parallel to changes in the minimum wage or the EITC. And so we throw this at it and the minimum wage results are fully robust to it. 

 

Michael [00:49:31] You can't really do a super high ordered time trend for the EITC because we have to do it in three year blocks. What you're going to find in general is there's an entire literature of really important robustness checks that we can do for the minimum wage. We don't have access to as many options because our EITC policy changes are so much chunkier and our outcome variable is again, it's so much coarser, so much chunkier. So getting back into it, we're going to throw what we call leading versions of the minimum wage. So like, we're going to put the future minimum wage as what we call placebo tests. And I'm pulling this from Meer and West's 2015 paper to make sure that there isn't some policy environment, right, that's driving this - that just happens to be corresponding to the kind of environment where you get a minimum wage. Again, the results are - in fact, I'm going to say in advance, the results are robust to all these things I'm about to discuss. We control for census divisions, and we do that because there could be a labor market that exists outside of the state. So the U.S. labor market is in many ways regional, not just state or local. So we throw controls in for that and there's no change. We steal something from Clemens and Wither 2016 where what if we're concerned that again, there's this policy culture within a state that is parallel in the labor market policy and criminal justice policy? So if that's the case, we maybe can't rely on as much on within state changes. So in that case, or at least the between state changes are so strong that they're going to pollute our results. So what we do is we look at only federal changes. There's three years of federal changes in our data. And again, we find the exact same result. 

 

Michael [00:51:19] Meer and West also expressed in the 2015 paper concerned that there could be delayed employment effects that for the minimum wage's disemployment to kick in because it shows up less in terminations and more in rehiring. So if you run a McDonald's and the minimum wage get hikes, you actually don't have to fire anyone. Your turnover is so high that when people quit as they just naturally do because it's a frequently temporary job for people, you just fail to rehire. So what that means is the effect is going to be lagged. So we we throw those in or we test just the lags and everything tells the same story. And then to pull a paper from Allegreto and Dube and other coauthors, I can't list them all. They have a paper in 2017 that - it got a lot of attention where they do a county border study, where they compare labor markets across borders that have different minimum wages. So what we do is we create essentially clusters of counties that share borders. And then we use, we don't know the county you live in in our data. We don't know the county our subject's live in. But we do know the county for the court that sentenced them, and that's not a completely crazy proxy for the county they're likely to live in and we perform that analysis and we generate the same results. It's really tempting to highlight those results, but because we have so much less faith that we actually know the county you live in with enough accuracy, we still think the state level results are a lot more reliable. So that's what we leverage. But you can dig our appendix. You can find the county border study. We look at the scrap in the minimum wage literature and we try to take specification choices from papers in the literature that get countervailing results and we pull tests from both sides. And the nice thing for our results is that whether you take specifications from the literature that says the minimum wage matters and has negative employment effects are the ones that says it doesn't, you can take either their specifications and generate the same results on recidivism in our paper. 

 

Jennifer [00:53:30] Okay, great. So finally, one more check that we should talk about. You use data from the current population survey to consider effects on employment for those most likely to have criminal records. So first, why do you do this? And second, what do you find? 

 

Michael [00:53:45] So we're really curious about the underlying employment effects, but were foiled because there's no employment data in our prison data. So this is one of those great coauthor moments. So this is Amanda's idea, and we're just chatting one day and she just throws this out at me and, you know, we were on the phone, but I could have hugged her. She's like, you know, in the CPS, there's a voting supplement and there's only 5,000 people in it, but we do know if they've lost their voting rights. So - and in the U.S., if you're a citizen and you have been disenfranchized, it's almost exclusively because you have a felony record and you live in a state that doesn't let you vote if you have a felony record. And it's been estimated that one in three black men in the U.S. carry a felony conviction record. So if we look at the voting supplement, we find that a one percent minimum wage increase actually increases the employment of people who are disenfranchized by .37 percent, which is a small elasticity, but again, not trivial. And more importantly, it's positive. You know, we're, you know - the entire point of Econ 101 is there's no reasonable mechanism by which the minimum wage should increase employment in a competitive market. But that's exactly what we're seeing for people who are disenfranchized, which would support our theory that the minimum wage if you have a felony record is actually pulling you into the legal labor market. I mean, again, this is the current population study and we're looking at a voting supplement, but we're actually seeing support for our employment mechanism. And then to kind of push that story farther, about one in three black men in the U.S. carries a felony conviction record. 

 

Michael [00:55:35] If we stratify the employment analysis of the minimum wage to just black men without post-secondary education in the U.S. between the ages of 18 and 54, so this is going to be the predominant - this is going to be the largest subgroup with a felony conviction. We actually see that a one percent minimum wage corresponds to a .16 increase employment. Again, it's a small increase. But the fact that it's an increase at all, you know, offers credence to the story that while it's completely reasonable to expect net disemployment effects in the minimum wage for the broader population, for people carrying this criminal stigma who are facing just completely different labor market choices. And most importantly, and I haven't really heard anyone else speak to this, once you have a criminal record, that first arrest, that first prison sentence, radically changes your labor market earnings for the rest of your life. The relative cost of the second, third, fourth conviction are so much smaller that it's just - there's not the same opportunity costs, so you face a completely different decision between legal and the illegal labor market. So again, there is this opportunity for the wage effect to dominate your decision making once you have a criminal record. 

 

Jennifer [00:56:58] Right. And just to reiterate, the story that you're telling in the paper is that, you know, you have this outside option. You could work in the illegal market and sell drugs or sell stolen goods or whatever it is and make a lot more money, potentially. And so if the minimum wage increases, it pulls some of those maybe even like you could describe as higher skilled criminals back into the legal labor market. And so you're basically finding that that effect seems to dominate, which is also relevant to the other mechanism I throwed out at the beginning that a lot of these people are employed anyway, and now their wages are increasing. And so maybe those are the people that are committing less crime and the fact that you're seeing a positive employment effect here suggests that that might still be part of the story, but there is also this change in the composition of workers, so it's not that - can't be the only story.

 

Michael [00:57:47] Yeah. And you know, a lot of - in my priors on the criminal labor market is that there are barriers to entry and there is human capital in that market, meaning their skills. You can't just know that you want to steal things. You have to know how to slide under an SUV and pull out the catalytic converter. You need to know how to sell stolen goods in the secondary market. You need to know how to find a wholesale drug provider so you can become a drug retailer. You need to know how to participate in the illegal sex work market. And once you're in the criminal system and for lack of a better term, the broader criminal community, you're more likely to have those skills and connections and not have those same barriers to entry. And then you combine that with the lower opportunity cost in terms of your, you know, potential labor market earnings, future destruction, it all points in the same direction and whether it's the EITC or the minimum wage, there is some range of possibility where if we push those lower skilled, lower earning markets up, they can become more competitive with the illegal market and those sort of grayish markets. 

 

Jennifer [00:59:00] Okay, so when you compare, head to head comparison here of the minimum wage versus the EITC like a $1 increase in the minimum wage versus a $1 increase in the EITC, which do you find has a bigger effect on recidivism? 

 

Michael [00:59:13] Well, for men, as it stands, we're very comfortable saying the minimum wage has a bigger effect. 

 

Jennifer [00:59:20] Because the EITC doesn't apply to them, basically. 

 

Michael [00:59:22] For most of them, they're staring down an additional maybe $500 a year, and that's probably just not going to be enough to move them. And it doesn't appear that it is. I think what we would be willing to say is your standard EITC introduction has a bigger effect than your standard minimum wage increase or your average minimum wage increase. And I think we're also willing to say that there's less reason to expect an eventual overtaking of a negative effect with - there's no anticipated future negative effect for the EITC. So we don't have to stare at the horizon and go, oh, once the EITC hits 100 percent of the federal EITC, or once it hits $10,000 and once it hits $15,000, we'll see the effect turn around. There's no reason to expect that, whereas with the minimum wage, we have to think of this more as it's - there's more technocratic precision that we - there's more of a bullseye you need to hit. 

 

Jennifer [01:00:21] Yeah. Well, and even at the current level, I think your results are in line with the idea that there is some disemployment effect. And what you're finding is that the pulling people into the legal labor market effect is overtaking that for the minimum wage. But the EITC, everything's moving in the same direction. So you - so we would expect theoretically to see a bigger effect with the EITC and I think that's what you guys find. Okay, so that's your paper. Have any other papers related to this topic come out since you first started working on the study? 

 

Michael [01:00:51] Yeah, there's a couple of papers. I mean, I can't shout out every paper that's been about the minimum wage since we - then because there's one a week and I can't shout out every crime paper that is you know related to, maybe tangentially labor markets and crime. There's three papers I'm a really big fan of. So Dow, Godoy, and Lowenstein and Reich have a paper about how economic policies reduce suicides, what they call deaths of despair, and they find that the EITC and the minimum wage both reduce non-drug suicides, which is good because you could think - they're able to then separate sort of acts likely or potentially at least accidental suicides. And so it's further evidence that these labor market policies are changing the life prospects of people that these are - that even relatively small policy changes here are having really big effects on people's lives, not just for whether they go back to prison. 

 

Michael [01:01:45] There's a great paper that's really directly related by Galbiati, Ouss, and Philippe that looks at re-offending after incarceration and recidivism rates, essentially with slightly different data. And what they look at is new job advertising, and what they find is that when jobs are advertised more, that reduces recidivism that that - that just knowing about these jobs pulls people out of the illegal labor market. That's again, one of those things where it's hard to appreciate. I think for a lot of academics, just how low an information environment someone being released from prison operates in. They've been literally living in a information void and say what you will about the minimum wage. It is extremely well advertised. So people know what it is. So advertising makes people aware of where the jobs are. The minimum wage might make people understand what they're - what they could stand to earn in the legal market. 

 

Michael [01:02:42] My rule of thumb is if Jeff Clemens writes a paper, I want to read it. So he has a paper with Michael Strain that actually looks at noncompliance with the minimum wage, which is this kind of gray market paper. And this is not the fault of economists. This is the fault of just the world we live in. This is that the nature of data. We just know so little about the extra legal market, the gray market. And so what they find is that there's a lot of minimum wage noncompliance, which to me is, yes, is an example of wage theft. But it's further evidence that there is this entire sort of for lack of a better term underground economy that we just know very little about what people's options are within it. 

 

Jennifer [01:03:32] And by noncompliance, you mean being paid under the table or something like that? 

 

Michael [01:03:36] They're being - yeah, they're being paid a sub minimum wage or some portion of their wages being paid under the table. This is one of the things I think changes generation to generation. If you talk to an economist in your department over the age of 60, about what fraction of their earnings were cash under the table and you know, before the age of 25, a lot of times they'll tell you all of it. I mean, I had plenty of cash under the table earnings until I graduated college. If you've - anyone, anyone who's listening, who grew up in maybe a less than affluent environment understands that you don't take your car to a professional mechanic. They're incredibly expensive. You have a friend who has a driveway, you park your car there, you give them 50 bucks and they fix it. I didn't pay a legally employed person to fix my car until my 30s. So I mean, there's just and we won't even get into, you know what it means to fence an object or to participate in the market on Craigslist. So I mean, there's whole world's - whole labor market worlds that we just know nothing about. You know, I've referenced it three times. And let's do it a fourth. So in 1938, the U.S. enacted a minimum wage that was foisted upon Puerto Rico, and it was a couple of hundred percent larger than their median wage. And so the reported unemployment rate cracked 70 percent there. Now, do we actually believe that 70 percent of Puerto Ricans stopped working and earning an income and became fully autarkic Robinson Crusoe is living on an island. No, they just all started getting paid in cash. Everybody went off the books the next day. And that's not going to happen at the same magnitude with the kind of minimum wages the U.S. is typically, you know, meddling around with over the last 30 years. But that doesn't mean that this unobserved - just because we don't see this labor market doesn't mean it doesn't exist, and it isn't salient to the decisions being made, especially by people being released from prison. 

 

Jennifer [01:05:37] So what are the policy implications here, both of your paper and the other work in this area for any policymakers or practitioners or advocates out there? What should they be taking away from all this? 

 

Michael [01:05:48] Disconnect the EITC from child requirements and marriage requirements. Just fully separate. There's absolutely no reason for it. I mean, I understand politically there is, depending on the market within, it may be easier to sell giving money to families than people who are single. But the effects of the policy are so strong and they're so unequivocal that there's - it's worth expending political capital to remove those limitations, especially because at the end of the day, a politician knows that they have to sell policies. But people vote their checkbooks and people vote their crime rate. And if you have a policy that is going to increase people's incomes and lower the crime rate and lower recidivism, you should do it. It'll be in your interest. 

 

Michael [01:06:38] And then this is a more subtle take away. But I think this is a policy take away that everybody can learn from, which is that these are not big changes in the minimum wage. This is not enormous changes in your yearly tax refund. This is an extra dollar an hour. This is an extra, you know, $2,200 a year that are radically changing life trajectories, which means that any policy that can reliably put more disposable income in the hands of struggling families in the families that are - or households or individuals that are more likely to have to resort to extra legal income mechanisms, these policies that might feel trivial to someone who enjoys a middle class income and a college education and the privilege that comes with that these relatively small changes are not to be trivialized and shouldn't be dismissed. You don't have to dig your heels in and say all or nothing. That if we can't give an extra $20,000 a year to a household, it won't make a difference. It's just not true that it's worth fighting for even small changes. 

 

Michael [01:07:52] As for more speculative policy stuff, the employment options for people with criminal records are just so - they're just so thin. And any policy that leads to fewer people carrying a felony record that can walk away from an interaction with the criminal justice system without a felony record goes a long way to improve outcomes. There's a paper - I always want to plug the paper by Michael Mueller-Smith that looks at how, you know, policies that change, whether or not someone has a felony record when they leave the criminal justice via the probation mechanism. It could - changes their life trajectory going forward. And this isn't as directed a result as his is. But this is just one more piece of evidence in that bucket that says anything that we can do that can guide people through the system without a felony record is for the better. 

 

Michael [01:08:48] And then let's do more speculative that you definitely should not blame my coauthor for having because I don't know how she feels about this. But I mean, this is more evidence that things we can do that that legalize currently illegal mechanisms for generating income that don't have a lot of external damage can have the effect of pushing people back into the legal labor market. So legalizing marijuana, legalizing sex work or just similarly criminalized income opportunities and not necessarily even legalized, but just decriminalizing so that people who have to resort to those mechanisms don't have to be processed as a felon. That can go a long ways in the long run of pushing them back into the legal labor market, which doesn't just have the benefit of there being lower crime, but it has the benefit of meaning that they participate now in a labor market where they enjoy legal protections. I mean, life is different in the legal labor market. 

 

Jennifer [01:09:44] And what is the research frontier? What are the next big questions in this area that you and others are going to be thinking about going forward? 

 

Michael [01:09:52] Anyone who's been listening for this long knows that I am just deeply curious about what can we learn about the criminal and extra legal and gray labor markets? We just know so little about what they earn, what their lives are like, the hours they work. Do they have any sort of mutual protection systems that because they can't, you know, rely on the law to protect them, do they have other mechanisms? Are these criminal earnings dominant? Are these their principal sources or are these largely just side hustles? And that's maybe because their side hustles it's why small policy changes can have such big effects. And what it lends itself to is any grad student who wants to or any researcher who wants to put resources towards essentially quantitative ethnography where we can delve deep into these roles and actually learn what these labor markets look like. You know, if anybody wants to put up the Dave Skarbek bat signal and say, hey, you've taught us a lot about prisons, any chance you want to teach us what these labor markets look like? Maybe he's training graduate students at Brown right now, and he can, you know, steer them towards this and put me in the acknowledgments. That would be great. You know, that's what I'm looking for out there, and that's the sort of data I'm looking for. I mean, personally, I'm starting to get more into the criminal justice processing side and what it means to go through the criminal justice system, specifically as a person who earns income illegally. But I'm like a lot of economists, I'm guided by my curiosity, but I don't jump in fully until I know where the data is going to come from. 

 

Jennifer [01:11:37] My guest today has been Mike Makowsky from Clemson University. Mike, thank you so much for talking with me. 

 

Michael [01:11:42] This has been an absolute treat. Thanks for having me. 

 

Jennifer [01:11:50] You can find links to all the research we discuss 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 enjoyed 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.