Episode 56: Gaurav Khanna

 
Ep56_Khanna_photo.jpeg

Gaurav Khanna

Gaurav Khanna is an Assistant Professor of Economics at UC San Diego's School of Global Policy and Strategy.

Date: August 17, 2021

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

A transcript of this episode is available here.

Episode Details:

In this episode, we discuss Prof. Khanna's work on how formal employment affects criminal behavior:

“Formal Employment and Organized Crime: Regression Discontinuity Evidence from Colombia” by Gaurav Khanna, Carlos Medina, Anant Nyshadham, and Jorge Tamayo.

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 Gaurav Khanna. Gaurav is an Assistant Professor of Economics at UC San Diego's School of Global Policy and Strategy. Gaurav, welcome to the show.

 

Gaurav [00:00:28] Thanks, Jen. Delighted to be here.

 

Jennifer [00:00:29] Today, we're going to talk about your research on how disincentivizing formal sector work in Colombia affects criminal activity. But before we get into that, could you tell us about your research expertise and how you became interested in this topic?

 

Gaurav [00:00:42] Right. So, you know, I'm a labor economist, which means I mostly focus on topics related to employment opportunities, human capital, migration, conflict and, of course, crime. A lot of my research is focused on emerging economies in South Asia and Latin America, but also the U.S.. I first started working on conflict in South Asia and sub-Saharan Africa, where we focused on economic incentives that drive people to join rebel groups and militias like, for instance, related to the Maoist insurgencies in South Asia. And, you know, while I was working on these issues in grad school in 2016, I went to Medellín in Colombia for a conference. When I was there, some of these issues that I was studying elsewhere in the world just seemed so salient in this context. One of the biggest concerns plaguing cities in Latin America is that many of these youths are being drawn into a life of crime. And so when I was there, I really started talking to people about the data that they had and what kinds of issues that the youth in Medellín were dealing with to really try to understand more broadly why are the youths in such cities being drawn into crime.

 

Jennifer [00:02:00] So your paper is titled, "Formal Employment and Organized Crime: Regression Discontinuity Events from Colombia," and it's coauthored with Carlos Medina, Anant Nyshadham, and Jorge Tamayo. And you're focused on the context of Medellín, Colombia, specifically. So give us some background on organized crime in that city to help us understand the context.

 

Gaurav [00:02:20] Yeah, you know, so Medellín has this pretty depressing history with organized crime, really. It's a beautiful city. And I would encourage all of your listeners to go and visit if they haven't already. It's really an amazing city with a great art scene and a great music scene and amazing food and wonderful people. And it has a very interesting topography. It's in a steep valley with a river running through the base of the valley. And most of the people actually live along the steep hillside. But it's also a very segregated city. You know, there are entire hillsides that are just slums and there are entire hillsides that are big mansions. And, you know, the city is surrounded by forests, which makes it, again, very beautiful. But these forests had been used as hideouts for producing drugs at a pretty large scale—often the drugs actually that end up coming to the US are often from that region—and also the forest we use for kind of hiding militias, both right wing and left wing militias—and, you know, over time, these cartels and militias actually moved into the city neighborhoods, most famously under Pablo Escobar.

 

Gaurav [00:03:38] So, you know, while it's a beautiful city, it has this really violent past with extremely bloody gang wars, where cartels and militias would really try to fight over territory. And at the start of our study period, around 2000, the turn of the century, it was one of the most violent cities in the world, had among the highest homicide rates in the world. Things have gotten a lot better since then. But even a decade later, it was still one of the ten most violent cities, along with other cities in Latin America and Kandahar. And, you know, some of you may have seen the TV show Narcos set in Medellín, which I think unfortunately humanizes the drug lords a little too much for my preference, but otherwise describes violence in the city pretty well. And this history really means that there is you know, there's a lot of guns floating around the city and and there's enough youths who have had experience with how to use a gun. And so that even if police really crack down on certain types of cartels, many of these youths join militias. You know, they have experience with guns. And, you know, much of the cycle of violence really continued for decades, really.

 

Jennifer [00:04:52] And in the paper, you write about how gang members actively recruit young men to work with them so tell us about this process and the groups that they seem to be targeting.

 

Gaurav [00:05:01] Actually, some of the best descriptions of this recruitment process come from reading interviews with gang members that actually, you know, anthropologists and sociologists and ethnographers have been doing for a while. You know, I again encourage your listeners to actually read some of these interviews. I remember picking up a dissertation by actually an anthropology student, Adam Baird, who interviewed a lot of these gang members. And one of the things that really pops out at you when you're reading these interviews is how important these economic incentives are during this recruitment process. Right. They just- the economic incentives are really salient during this recruitment process. Essentially, what will happen is that gangs will target what they called "idle youths" in their neighborhoods. So youths that they know don't have a job or at least don't have a great job and, you know, approach them and say, hey, do you want to work for us or do you want to do this other normal job? And what this means is that, you know, those with good formal sector jobs are extorted but not really recruited. But then other youths were kind of hanging around the neighborhoods for most of the time. Well, they say they end up joining these gangs, you know, when you interview them—well, not me but when ethnographies interview them—they say, well, we joined because it was a really profitable career. Right.

 

Gaurav [00:06:27] And the industrial organization of these gangs makes it so that there are always recruiting. There are about 300 street gangs in the city and the competition is so fierce that essentially they really need to keep having youths join their side. Right. So they're always trying to kind of actively recruit. From the side of the youth, you know, it can be quite a lucrative job opportunity, right. You get promoted with time and over time you start making quite a lot of money. You know, gang leaders are among the richest folk in the city, actually, and they really run these gangs as profit seeking organizations. You know, just as enterprises. They own revenue from protection services, from extortion services and, of course, drug sales. So when you read this recruitment process, you know, these economic incentives really kind of are at the forefront of things here.

 

Jennifer [00:07:21] And we'll make sure to put links to some of those ethnographies and interviews you mentioned in the show notes because I'm sure people will be interested. So in the paper, you consider the effect of formal sector employment on criminal activity. And it might seem obvious to people that there's a link there, but it's not obvious to researchers or has been difficult to pin down. So why is this such a difficult question to answer?

 

Gaurav [00:07:44] Yeah, you know, so you might think that alternative economic opportunities are going to be quite important when folks are making decisions, whether I should join criminal enterprises or do other kinds of jobs, but like you said, it's it is a really challenging question to answer. Right. Especially when you are interested in how individual level incentives for these youths really drive whether they choose a life of crime or not. Right. Now, because crime is such a rare outcome, often we realized when we started this project that we needed at least two things to find good causal evidence that kind of leverages this individual level variation. First, we needed individual level administrative data on the universe of crime related outcomes, including, you know, what kinds of crimes they were committing and when they did it and so on and so forth. And second, we needed individual level, quote unquote, "exogenous variation" in the returns to crime or the returns to formal sector employment. So really changing the incentives and the costs and benefits of being involved in crime or informal sector jobs.

 

Gaurav [00:08:54] Now, researchers in other settings, especially in northern Europe, for instance, have done an excellent job of putting together these kinds of individual level data. And now we saw these and we were really inspired by them and said, well, you know, we'd like to do this in this very high crime setting, in this low income setting of Medellín where about one fifth of these young men in our sample at some point in time were arrested during their kind of criminal careers. And so what we did was we started putting together this kind of administrative data and all credit goes to my enterprising coauthors who got all kinds of permissions from local police, local city officials, to get administrative records that really allowed us to match individuals across databases. So, you know, it's really time consuming. This process took us about four years and is continuing in many ways for kind our follow up papers, but that was, you know, getting the data- the measurement itself was really challenging for us here.

 

Jennifer [00:09:57] Yeah, it's super interesting to see this kind of data collection and putting together of these kinds of administrative records in Latin America. It's been a relatively recent development and I'm super excited about it. So as you said, this is a somewhat different context, especially from places like Sweden.

 

Gaurav [00:10:12] Right. And, you know, and so far, we've not just matched like crime records to kind of household censuses, but we've brought in like foreign based censuses, credit histories, school based records, you know, who you were in a classroom with, records of various social programs. Basically, you put together this kind of treasure trove and we said, well, now we can dive into it. It took us a long time for us to dive into it to try to answer this question, why are youths in Medellín getting involved in crime?

 

Jennifer [00:10:39] Yeah. So, yeah. So you've been working on this topic for a while. So before this paper, what had we known about how employment affects crime?

 

Gaurav [00:10:49] Well, you know, there's really a long history of trying to understand how employment affects crime and people talk about how Gary Becker and Isaac Erlich had these models in the 60s and 70s talking about how individuals are choosing between crime and legitimate forms of employment. Now to kind of test these theories of individual behavior in a causal manner, as researchers, we often started relying on what we call area based variation—so variation that came from local unemployment rates or local recessions or large scale plant closings or trade shocks, a variation that adversely affected employment opportunities for an entire county or large neighborhoods or the entire city in some ways. Now, this area based variation is very interesting, but it was essentially answering a slightly different question because when there are local recessions, there are other general equilibrium consequences. It's not just changing job opportunities. There'll also be less resources for policing. There'll be less resources for gangs to expropriate or extort. There'll be changes in local demographics. You know, they might be white flight like there was in the U.S. So kind of area based variation is interesting, but it is a slightly different kind of variation, really. And this is a point made by Dix-Carneiro, Soares and Ulyssea in their paper on trade shocks in Brazil that really it's a very careful point that's saying that, look, it's a slightly different elasticity altogether. Right.

 

Gaurav [00:12:30] But the reason we relied on area variation is that finding good causal individual level variation is really challenging. Right. More recently, researchers have done a good job of doing this. You know, they often in settings in northern Europe and also in the U.S., one of my favorite papers is this beautiful paper by Paolo Pinotti on how work restrictions for immigrants to Italy really affected their involvement in crime. And, you know, that paper really showcases this the kind of the power of individual level administrative data and kind of deriving this individual level variation. Well, we saw these developments in the literature and said, well, can we also do this in Medellín? So that was kind of our goal. We wanted to really derive this individual level variation in employment opportunities to get at how that then affects whether youth get involved in crime.

 

Jennifer [00:13:18] You're going to use a rule about eligibility for subsidized health benefits as a policy shock that disincentivized formal sector employment as a way to measure the effect of formal sector employment on crime. So tell us about how health benefits work in Colombia.

 

Gaurav [00:13:34] Yeah, so you know, health benefits and how they're actually paid for are really salient issues in Colombia. And they're also kind of a political hot topic. So there's two types of benefit regimes. First, for anyone in the formal sector or for anyone above a certain income level, they are in what we call the contributive regime, which means these folks like pay into the system, whereas low income, non formal individuals are in what we call the subsidized regime where they don't have to pay anything into the system. Right. So essentially, it's the richer or the formal workers that are cross subsidizing the non formal, low income individuals. Now, earlier, there were some differences in the coverage—you know, the health coverage was slightly different as well—but the Supreme Court eventually ruled that they needed to cover the same set of things. And by the end of our study period, ninety six percent of the population was covered. So most people had health insurance of some sort, and more than half of them belonged to this kind of subsidized regime.

 

Gaurav [00:14:37] So basically, everyone had health coverage, but there was a difference in who paid for it. So richer or formal sector workers would pay up to twelve and a half percent of their wages for this coverage. And some of this money would go into covering the non formal, low income workers. So what about the non formal, low income workers? Well, they didn't have to pay, but to not have to pay, you have to meet these two criteria. You have to be below a poverty score cutoff. So you have to be poor. And secondly, you had to not be in the formal sector. Right. And it was the second criteria that people worried could lead to distortions. Right. So, you know, academics and the media and politicians also have argued that this meant that certain households that were below this poverty score cutoff may actually not try to join the formal sector simply so that they'd then be eligible for these benefits for free, essentially.

 

Jennifer [00:15:41] Yeah. So they're actually better off if they stay out of the formal sector.

 

Gaurav [00:15:45] That's right.

 

Jennifer [00:15:45] That gives them a strong incentive. And so I gather there had been previous work showing that this program did indeed disincentivize formal sector employment, so tell us about that research and the findings there.

 

Gaurav [00:15:56] Yeah, there's really some excellent work on this, both in English and Spanish, actually. And, you know, they're using different identification strategies and research designs. For instance, one of my favorite papers by Camacho, Conover, and Hoyos, what they do is they leverage the roll out of this program across different municipalities in Colombia, and they really show how these benefits reduce formal sector employment by about four percentage points. So there's some other work by Gaviria and coauthors and --- and coauthors so this- there's a lot of research really showing how this program disincentivized formal sector employment. And after a while, the government actually started recognizing that, yeah, there were a bunch of these kind of fees and taxes that low wage formal sector workers were paying, such as this fee that was really disincentivizing formal sector employment. And, you know, there's actually some great research by Adriana and Maurice Kugler that was kind of instrumental in making this point. So it became a big political issue. And then, you know, at the end of a study period, there were actually new laws that were passed that then said, OK, we recognize this is quite a problem, actually. So we're going to lower these fees that low wage, formal sector workers are paying to really try to encourage formalization.

 

Gaurav [00:17:18] And then, you know, economists, we never give up an opportunity to write more papers. So, you know, there was another a research paper that showed that when this law change happened in 2013, this actually led to an increase in formal sector employment. So, again, the- Adriana and Maurice Kugler wrote a few papers on that. So you know, really there's a lot of research that shows that making low wage, formal sector workers pay so much into the system disincentives formal sector work. And then when you took away those payments, when you didn't ask them to pay, it actually led to an increase in formal sector employment. So there's a really good research agenda out there.

 

Jennifer [00:17:56] So people respond to incentives, turns out.

 

Gaurav [00:18:00] Surprising.

 

Jennifer [00:18:01] So you're going to use the same cutoff to consider the effects on criminal activity and particularly criminal activity that appears to be related to organized crime. So walk us through your empirical approach. How do you measure the causal effect of this disincentive on criminal activity?

 

Gaurav [00:18:16] All these great researchers have already shown how these disincentives discourage formal sector employment. And so where we came in was, well, if they're not going to join the formal sector, what are they going to do? And in Medellín, you know, a very lucrative option outside the formal sector is actually join a criminal enterprise. Right. So, like I said, our goal was to derive variation in the costs and benefits of being formally employed. Right. And the policy rules were such that it actually ensures that the costs of being formerly employed jump discontinuously at this poverty score cutoff that I mentioned. So let's talk a bit for what is this poverty score cutoff? Right.

 

Gaurav [00:19:04] So the Colombian government actually does a household census and measures all kinds of things when they survey these households, like how many rooms you have and what kind of assets you have. And they use a secret formula to then create what they call a poverty score or the Sisben score. And then they use this Sisben score to target different social programs at different values of the score. And at one such value, they have this cutoff below which if you're if you're below this value, then you're poor. And so you kind of get these health benefits for free. As long as you're not in the formal sector. Right. So households on either side of this poverty score kind of are almost identical, except that maybe one of them may have one less asset or very slightly lower level of income. And so suddenly the household that has slightly less income, suddenly becomes eligible for free benefits if they drop out of the formal sector. Right. So we can then compare these almost identical households where the only difference is the kind of these costs of accessing these benefits and then evaluate whether this difference in these costs of assessing benefits actually lowers formal sector employment–so did that households actually drop out of the formal sector or not join the formal sector in the first place? And then what do they do when they were not in the formal sector? Right. Did it increase participation in criminal enterprises?

 

Gaurav [00:20:34] So like I said, earlier research already showed that folks did drop out of the formal sector. And we said, well, if these youth are not joining the formal sector, what are they doing? And gangs will approach such youths that are not in the formal sector, try to get them to join—we talked about the recruitment process—and it's a really lucrative opportunity, you know, non formal opportunity for these especially young men. But, you know, there's something I want to highlight here about this poverty score cutoff is that there's actually something different about this variation that's quite different from the variation that I mentioned earlier about variation from local recessions or trade shocks. That variation on local recession compares different areas, like different counties or different municipalities, you know, one that's doing poorly and the other one that might be doing relatively better. Whereas here we're using kind of individual level variation, by which I mean we are comparing two households on two different sides of the Sisben score cut off and these two households may you know, they may even be neighbors. Right. So they may even be kind of down the street from each other. But the only difference is one of them becomes eligible for free benefits if they drop out of the formal sector. And so that's kind of the variation we're really using in this context.

 

Jennifer [00:21:47] Yeah. And that reminds me, actually, one more thing that I found fascinating- it's just sort of one additional fact about how this works, is that this is all measured at the household level. So you have these like multiple generations living within a household and you have a young person coming of age to the point where they're thinking about getting a job and suddenly everyone's health benefits are on the line if they go get a formal sector job. Am I remembering that right?

 

Gaurav [00:22:09] Yes, that's right. And you know, again, the interviews that I mentioned become quite interesting to look at. You see how often a lot of these decisions are like family, almost family based decisions, really. And then we have some other research- we have another paper that shows how there are kind of spillovers within households. So, for instance, if your family member loses their job, you actually see how that affects the other youth in the household, because it's such a closely, tightly knit group that there're kind of these other dynamics that happen within the house that are also quite important.

 

Jennifer [00:22:43] OK, so tell us about all of the amazing data that you have that you're using for this paper.

 

Gaurav [00:22:49] So my very enterprising coauthors, Jorge Tamayo and Carlos Medina, they're from Medellín and they actually convinced the police for us to share their arrest records going back to 2002. You know, these data are really great. They have, among other things, the exact article under the penal code, the exact article under which the arrest was made. They have the geolocation of the crime. And for a subsample of the data, the police have also flagged whether they suspect the individual belonged to a gang or a criminal enterprise. And then what type of gang, what is the name of the gang or the criminal enterprise, really. And importantly, for each arrest, they record their national identification number, which is kind of like the Social Security number and their year of birth. Right. So we kind of use this identification, you know the Social Security numbers and year of birth, to match individuals to Sisben censuses, which again also have these national ID numbers and years of birth, which again, you know, my very enterprising coauthors managed to obtain the Sisben censuses as well, that then they could match these data. And, you know, this is the same Sisben censuses that I mentioned earlier which document their assets and what social programs they have access to. But there's something really interesting about the crime data that I should mention now.

 

Gaurav [00:24:14] So we wanted to understand how incentives change your occupation, the kind of a job that you do. So we're thinking of being involved with criminal enterprises as kind of an occupational choice, almost. Right. So to do that, we needed to classify crimes into crimes that were what we call "likely associated with criminal enterprises" and crimes that were, you know, more crimes of impulse or passion or opportunity. And here is where that subsample of data becomes useful. The subsample where the police flagged whether the arrested individual was a part of an organized entity or not. Right. Because we can then see what types of crimes are more likely to be committed by such flagged individuals and then classify those crimes as being likely associated with criminal enterprises. Right. We use another method as well. Because we know the geolocation of the crimes, we can also see which crimes are more likely to show up in gang neighborhoods and you get a very similar list of criminal enterprise related crimes. But I found this classification pretty interesting, right? I don't know if there's other work out there that really does use this kind of data to classify crimes and the crimes that are likely associated with criminal enterprises and other kinds of crimes.

 

Gaurav [00:25:31] So at the end of the day, you find that crimes that are likely associated with criminal enterprises ended up being crimes such as drug trafficking, extortion, homicides, intimidation and so on. Whereas other kinds of crimes like smoking marijuana or domestic violence or receiving bribes or identity theft, those were not really classified as being likely associated with criminal enterprises. So we kind of have this distinction so that then we can test more clearly well, did the changes in formal sector incentives really affect whether individuals get more involved in crimes that were likely associated with enterprise activity, really?

 

Jennifer [00:26:12] And who is in your main sample?

 

Gaurav [00:26:15] We focus our main sample on to be on young men. These are individuals who are most likely to be recruited into gangs and and in their youth, they're essentially making this decision. Right. Should I join kind of try to get a formal sector job or not? And so we kind of focus most of our analysis on young men. About 62 percent of all the first arrests in our sample happen to individuals before the age of 27. If they ever show up in the arrest records, they're most likely to first show up in their youth. So all the individuals may have been arrested in their youth, but because our data starts only in 2002, we may not have their entire youth criminal history. Right. So we kind of need to focus really on folks that we can at least, you know, see where they were, at least when they were 12 or 13 years of age. Right. We also restrict our sample to be first arrests. And that's really because, repeat arrests depend on this kind of possibly endogenous length of sentencing. And so we kind of focus on first arrest. But it turns out this doesn't really seem to affect the results in any way, even if we include repeat arrests. So that's kind of who is in our main samples, the first arrest for young men.

 

Jennifer [00:27:39] And yeah it does seem like that's the group where, again, if you're thinking of this as an occupational choice, you're deciding which career path to choose, that is the age and the group that would be of most interest. OK, so you've got all this amazing data. You're focused on these young men. So what are the outcome measures you're going to be most interested in?

 

Gaurav [00:27:58] Right. Well, first, we just need to establish, you know, did this poverty score cutoff, did the Sisben cutoff lead to any discontinuities, any jumps in the uptake of this subsidized health benefit that I was talking about? Right. So first, we just need to establish that. Then we need to establish well, did that correspond- did this take up of the health benefits of this cutoff correspond with fewer individuals being formally employed? Right. And then finally, if your individuals who formerly employed, did it actually increase the likelihood that they would get involved in criminal enterprises? So those are the three main outcome measures we are focusing on.

 

Jennifer [00:28:42] Great. OK, well, let's go through each of those sets of results in turn. So first, as you said, you're going to consider the effect of that Sisben score in 2002 on eligibility for subsidized health benefits. So what do you find?

 

Gaurav [00:28:56] We find that, as one would expect, at the cutoff, there was a sharp increase in enrollment in the subsidized health benefits regime of about 26 percentage points. This is not surprising. People had already in some ways documented this. It doesn't go from zero to one hundred percent because remember, you could be poor, but you could still choose to be formally employed. So then you won't get the subsidized health benefit. But you do see the discontinuity at that Sisben score cutoff where there's a jump of 26 percentage points in being enrolled in the subsidized health benefit.

 

Jennifer [00:29:34] OK, and then next, you consider the effect of eligibility for subsidized benefits, again using that 2002 cutoff on formal employment later in 2009. So what do you find there?

 

Gaurav [00:29:46] Right. So then we find that this enrollment in subsidized health benefits coincided with lower formal sector employment, interestingly of about four percentage points, which, like I said, interestingly, is actually the same magnitude that I mentioned earlier that other researchers had estimated using different data and different identification strategies, really. So, again, this result is not new. Other researche has shown that enrollment in this subsidized health benefit would lower formal sector employment. And in some ways, we kind of replicate previous research to show that even in our data that there was this decrease in formal sector employment by four percentage points.

 

Jennifer [00:30:29] And finally, your main result, you consider the effect of eligibility for subsidized benefits in 2002 on their subsequent criminal activity. So what do you find?

 

Gaurav [00:30:39] Right, so this is the new result, right? So this is something that other researchers haven't, in this context, haven't looked at. So we find that eligibility for these subsidized benefits actually led to an increase in being arrested for crimes that were likely associated with criminal enterprises. So these are crimes like homicides and drug trafficking, extortion. But there was no change in arrests for crimes that were crimes of impulse or passion or opportunity. So these are crimes like domestic violence or smoking marijuana, so there was no change in arrest rates for those kinds of crimes. But for crimes that were more likely associated with criminal enterprises, including violent crime, property crime, drug crime, you know, there was an increase in being arrested for these kinds of crimes.

 

Jennifer [00:31:34] And comparing these effects on the criminal enterprise and non criminal enterprise crimes does help you think through potential mechanisms a bit. And in particular, it helps you rule out some channels through which access to subsidized health benefits might affect crime rates. So what are the potential mechanisms you have in mind and what do your results tell you about what's driving those results?

 

Gaurav [00:31:55] Yeah, yeah, you're totally right that it is really important to kind of use this distinction between different types of crimes to kind of rule out these other stories. Right. So, for instance, first, we may be concerned that formal sector workers who were paying into the health system were worried that if they lose their jobs because they get arrested, then they lose all their kind of vested benefits. Right. So maybe they were just being extra careful because they've already paid so much into the system and they didn't want to get arrested. So they didn't want to get involved with any kinds of other activities. So it could be that maybe that's all that's going on here because they've already invested so much in the system. Alternatively. Secondly, it could be that police may be falsely targeting non formal youth who were just hanging around the neighborhoods and going around and trying to pick them up and arrest them. Right. So they just see that they're not at a job right now. So they'll go and just try to arrest them for something. Or thirdly, it could be that the health benefit itself may increase the amount of risky behavior, for instance like consuming more drugs, that you are that you are now insured against because you have this kind of- you're covered by this health benefit. Right.

 

Gaurav [00:33:17] As a side note I would say that almost everyone had health benefits at some point. So the main difference was not about who was covered, but it was really about who was paid if you remember. You still want to be concerned about these other stories that could be happening in some way or the other. Right. But if any of these other stories were happening, it should really affect all kinds of crime, not just crimes that are enterprise related crime. Right. So if police are just rounding up folks that are hanging around or smoke or smoking marijuana or if people are not going to be engaging in more risky behavior like smoking marijuana or getting into fights then-.

 

Jennifer [00:33:56] Or pickpocketing or any other kind of like petty theft. Yeah.

 

Gaurav [00:33:58] Yeah, petty theft. So petty theft is a little complicated. If you're doing petty theft in a gang neighborhood, you need to be involved with the gang. But, you know, so seeing that there was no change in increase in arrests for other kinds of crime really helped us get more confidence in that what we were looking at here in the data was was really kind of an occupational choice, right. Really how these kind of formal sector disincentives created an occupation choice drove some of these youth into these criminal enterprises.

 

Jennifer [00:34:32] Yeah, I think that piece is really interesting and really compelling because it is like any other story you could think of, you would expect all the crimes to be affected and you really just see nothing going on for those other crimes. So I know that you have other ongoing work in this area. So what other papers related to this topic by you or anyone else have come out since you first started working on this study?

 

Gaurav [00:34:53] Yeah, you know, we've been working on this, like I said, you know, since we put together took a- it took us four years or maybe even longer to put together some of these data. We said like, well, there's so many interesting papers to write about this. Right. So, again, really surrounding how these economic incentives drive youth into this kind of life of crime in Medellín. For instance, we have this other paper that leverages mass layoffs at firms. So if you're say formal sector enterprise is shutting down for some reason, then everyone gets laid off. You can really follow because of the data we put together. You can really follow individuals over time because we know who everyone's coworkers we can actually follow individuals over time and see, well, are they likely to get involved in crime down the line. And then we can parse out, well, what types of folks are more likely to get involved in crime? And we can see what really mitigates these channels. So, for instance, we show how access to credit, both consumption credit, but also credit for starting your own little microenterprise, actually mitigates some of these adverse effects of being laid off.

 

Gaurav [00:36:01] But then you can also look at things like spillovers. You know, I mentioned earlier there were these spillovers across family members. So if there's an adult in your household got laid off and lost their jobs, actually, the young kids are also more likely to get involved in crime. So the effects of these layoffs are more far reaching in some ways when it affects other kids in the household as well. That paper is also out.

 

Gaurav [00:36:26] But now we're working on more issues related to things like segregation across neighborhoods and how transportation infrastructure really affects access to jobs in different parts of the city and how kind of the expansion of the transportation network over the last few decades really helped reduce the amount of segregation of opportunities and helped improve access to good economic opportunities on the parts of the city and actually led to reductions in crime as a result. So we're trying to use this research to look at some of these other issues.

 

Gaurav [00:37:00] But, you know, there's also been so much excellent recent work by other researchers in this area in the last few years. I already mentioned Paolo Pinotti and Diego Britto and Breno Sampaio have- do something very similar to our paper on mass layoffs. But they do it in the context of Brazil. Again, looking at how being laid off from your job actually increases the likelihood of being involved in crime down the line.

 

Gaurav [00:37:23] And then, you know, in the Medellín context, there's some really fascinating work by others, Santiago Tobón and Chris Blattman. And they have a- nice big army of researchers- have some very interesting research on gangs in Medellín right now where they really look at what role these gangs are actually playing in replacing the state. You know and providing services and taxing residents in view of the state, really. So they actually went out there and interviewed all these gang leaders and they convinced the state to actually randomly increase governance in certain neighborhoods to look at really how gangs respond. It's really fascinating work.

 

Gaurav [00:38:03] And this other research at this final research agenda started by Mica Sviatschi at Princeton, which is about crime in Latin America, more broadly—again, I encourage your listeners to look into that. She's worked with some really great grad students like Nikita Melnikov and Carlos Schmidt-Padilla, who we should all keep an eye out for as the next generation of top crime scholars, I'm sure. But this agenda they started is really fascinating. And it looks at, you know, entire criminal life parts. It looks at how gangs restrict mobility across different territories in the city, which then really affects economic activities in different parts of the city and and how competition between gangs really affects kind of how extortion payments play out. It's really eye opening stuff. I can barely keep up. But I learn so much from it really.

 

Jennifer [00:38:53] Yeah, so Mica came on the show a little while ago to talk about that mobility paper. So I will put a link to that in the show notes, I agree. All of this is just so fascinating. We become less and less junior every year. It's wonderful to see all of these incredibly clever, smart, and energetic people coming into the crime space doing all this very interesting work. So what are the policy implications of the main paper that we talked about and all the other work in this area? What should policymakers listening to this take away from all these results?

 

Gaurav [00:39:23] Right. So coming back to kind of the disincentives for formal benefits paper, you know, that we we mostly talked about. For the specific policy, the main takeaway really should be that we shouldn't disincentivize formal sector employment. For instance, you could just target these health benefits to low income households regardless of their employment status. Right. Regardless of whether they were formal or informal. So if you take away this employment status criteria, then it doesn't distort incentives to be formally employed. Right. And actually, to be fair to the policymakers, they did recognize this. They did change their policies after kind of our study period. They did change the policies. They actually did end up lowering how much low wage, formal sector workers had to pay into the system. So that was good.

 

Gaurav [00:40:13] But, you know, speaking more broadly, I think in some ways these research agendas are really showing that in these kind of environments where criminal enterprises are such a lucrative employment opportunity, one of the best ways to fight crime may actually be to kind of help these youths get good jobs. And I mentioned this other paper we published on how job layoffs increase, whether they can get involved in crime or not. Well, the flip side of that is, well, if you have safety nets for laid off youths, you could really prevent them from getting involved in crimes and criminal enterprises when they get laid off. Right. So even though, of course, policing can be an effective strategy, it's really important to know that other than policing, there has to be a way to discourage youths from joining gangs in the first place and understanding the fact that access to employment opportunities is really one of the big reasons why youth might join gangs tells us that, well, to really stem the supply of youth to gangs, we need to kind of help them get access to good jobs in different parts of the city.

 

Jennifer [00:41:29] This is making me think about to what extent it's possible to undo some of these policy effects. So, are you able to see once the policy changed and there's no longer a disincentive, I imagine there's some path dependency for folks who had sort of already made the occupational choice to go into criminal enterprise instead of formal sector work. Do you see any of them coming back or are they already on a different path and they're not budging?

 

Gaurav [00:41:54] Yeah, you know, it's a great question. So in the early part of the data in the 2000s, sadly, what you see is if you get arrested once, you are likely to really get arrested again and not show up that often in other kinds of work in our data. That has been changing, there have been a lot of rehabilitation programs where the government has essentially said we don't want to be, that you made a mistake in your youth and we don't want to hold that against you for the rest of your life. That has been somewhat controversial because it's a bit like forgiving things that some of these youths were involved in. But there have been a bunch of young kids who are back in the formal sector. We were waiting for newer data to be able to look at some of these issues more clearly. So we have we don't have a good answer really to that. We want some more time to kind of look at more recent data to get at this empirically, more carefully.

 

Jennifer [00:42:51] Yeah, that makes sense. The bigger picture, that it's hard to get people off of the path and rehabilitate once they're already on it. It's certainly a problem we have in the US, too, so not unique to Colombia. So what's the research frontier? What are the next big questions in this area that you and others will be thinking about in the years ahead?

 

Gaurav [00:43:08] Oh, gosh. You know, yeah, I feel like you're better suited in some ways to answer this question. I mean you must have talked to so many people in your podcast and elsewhere, the seminars that you organized. To me, you know—like you said, you had Mica Sviatschi on your podcast—you know, things that she's doing on criminal territories and competition across gangs is really inspiring. Or what Chris Blattman and Santiago Tobón are doing in terms of the kind of industrial organization of these gangs, you know, really mapping out how organized crime is organized. Like I said, we're working on some new projects that can kind of on the economic geography side of things, trying to understand how not house- neighborhood segregation and access to economic opportunities on the parts of the city are really important in many different ways. And so how would kind of the expansion of this kind of transportation sector. So it's a very interesting topography in Medellín, but that means that the transportation sector is essentially these cable cars that go up the slopes of these hillsides. And so these cable cars were rolled out over the last two decades, really. And we can see as and when they rolled out how that really affects whether these youth now are more likely to go and work downtown rather than get involved in their own local street gang.

 

Gaurav [00:44:26] But all this research by Mica and Chris and Santiago, essentially what they're saying is that, you know, they're very careful in saying that, so what if this is a low income setting? And so what if it's a really high crime setting? We can still get really good data on gangs and criminal activity. Right. They're saying that measurement is really important. And folks in northern Europe and even in the US have showed us how they can get really good data. Well, we can do it in these contexts as well, because it's such an important issue in this context that we really need to get this good data and try to answer these sorts of questions, really.

 

Jennifer [00:45:00] Yeah, I completely agree. That's really inspiring. My guest today has been Gaurav Khanna from UC San Diego. Gaurav, thanks so much for talking with me.

 

Gaurav [00:45:08] Thanks, Jen. It was great chatting.

 

Jennifer [00:45:16] 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 Jon 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.

Jennifer DoleacRegression-Discontinuity, Employment