 So it's super great to talk off the chairs here. It's a very related topic. So I'm going to talk today about criminal enterprises from Rio de Janeiro, basically. So we are in an inequality conference. And we typically use favelas and slums to show, I think it's a clear portrait of the inequality we face across the world, in particular Latin America. And Rio de Janeiro is a city market in its landscapes for its favelas that have all the problems that we typically have in slums across the world, like segregation, problems of access to infrastructure, and poor social economic indicators. But in the case of Rio, on top of any all of this, we have a huge problem with violence. This graph shows the number of days in the city of Rio that you have at least one report from a civilian that there was a gunfight in a particular place of the Rio. So basically, it's at least 200 days per year. And as we see in this graph, mostly take places with favelas. And why this happened, the problem of the city is that we have several armored criminal groups that are located mainly in favelas, but not only. Some of these criminal groups are drug factions. Drug syndicates. The most famous one is Comando Vermelho, which was created inside the prison system in the 70s. But in the 80s, it starts to use the favelas as a place to sell cocaine. And because this is a super profitable market, this soon led to increasing disputes among gang members. And as a result, some gang members laughed and create two other groups. There are Teseiro, Comando and Amigos Los Amigos. And this fractionalization led to more battles over the control of the favelas and the increasing militarization of their gangs. So, but what is increasingly striking the city is that yet over the three days of escalation, the syndicates neither disappeared, nor adopted no violence strategies toward the state. And in addition to that, we also have in real police-linked militias. What is that? These are armored groups led by police and former police officers that began to dispute territories with drug gangs. In the beginning, they were viewed as a necessary evil, although because they would free the communities from drug dealers. They also dispute territory control with drug gangs and sometimes between them. And even, but they typically collude or partner with government officials to avoid repression. And what the police does in that context, oh, sorry, and before that, so militia groups has another particular characteristic. It mostly operate in protection markets, but are increasing accounts that they explored several markets in the city. They are involved in retail of gas canisters, illegal gambling, local transportation, and increasingly in land and construction. So, and if you go to the city nowadays, people will consider that this is one of the main threats for public security, and there is a widely idea that they are expanding a lot. But we know we've not much for the hard data on that. So what governments do, what the police do, say typically like they react to that with very violent military raids. So the police in Rio is not that they are not present. They typically do not patrol favelas, but they are intermittently involved in police incursions, super violent too, with like 100 of policemen heavily armed. And for you have an idea on how the police is violent in Rio. Last year, the police alone killed officially 1,300 people. This represents, it means that the police alone is responsible for an homicide rate of eight per 100,000 in the city. And this has been quite stable over years, but striking homicides committed by civilians has been decreased a lot in the city in the last 10 years, particularly in the last four years. And it's down to the lowest level that we have recorded. And so what we're gonna answer here, what we, the questions is, in order to understand this dynamic, to understand what they are doing, how they are expanded, first we need to have data on where they are. So we want to show where the armored groups operate, how they expanded over time, what are the incentives to criminal groups to engage in their force, when the state intervenes and why do some groups are able to increase its economic outreach. So the key challenge to do that and to understand criminal groups is to measure where they are. So what we do in this paper is that we use disc denuncia. Disc denuncia is a very important crime hotline in the city that there are more than one million reports over the last 15 years to disc denuncia, but we use this, typically people call the disc denuncia and describe what is going on and request for any type of help. So we're gonna use this disc denuncia and not just to try to track mentions of drug factions and militias, but actually we automatically interpret the content of each report and propose a rigorous definition of group present based on armored territorial control and exploration of economic activities. So pretty much what we do is that we try to classify the reports on whether they are reporting, such as these examples, they are heavily armored, they are exploring activities, they are carrying guns, charging monthly cash fees, and we are discarding all the reports that simply mention there is a militiano around someone and there is a funk party or whatever. So how we did that, so we have like we create regular expressions to classify the reports to mention whether they refer to armored territorial control that we define as armored circulation or roadblocks or any type of surveillance. And we have a second dimension which is the exploitation of economic activities which can be extortion or any charge of fees, illegal goods and service, and from drug trade to gambling to property crimes to also legal goods and service. So in a compliment to Alcelsia was telling that we have huge presence in listing markets there too. So basically, so we went from 1.4 million reports that we filter militian drug gangs name, we classify these reports, but then the crucial difficult here is that this is a civilian report that was not have been investigated and confirmed. So what we do in a nutshell is basically like, so this is a zoom of what is going on. So each dot is a report. So anytime it's colored, it's because it's these reports, they mentioned the Teseru comando puro, these the red ones mentioned comando vermelho and the black ones mentioned militia. So when you look a map like that, it's clear that Teseru comando is here, clear that militia is here. And some indication that survey is here, but I cannot take just one like that and say that there's a piece there. So what we do is that we aggregate this report, these dots by like creating clusters and whenever, and we did that for the whole period of analysis, so it's important to say we all opinion database go from 2008 to 2019 because we didn't want to include the pandemic period. So anytime in a cluster like that, we have a mention of the group name plus two mentions of the armored territory control and two mentions of exploitation of economic activity. We mark as the evidence that the group is present there. So it's a very strict definition. So we really want to get rid of false positives. So we really want to say that when the group is there, the group, it's really, we have good evidence on that. So we can do that for the whole metropolitan region. And what we want to track is this kind of phenomenon. So this year here in this neighborhood in 2009, we have indications of Comando Vermelho, Milicia and TCP. Then 2014, Comando Vermelho and Milicia and then the Milicia consolidated it control. So this goes to the panel data set as dummies that indicate whether you have more than one group, two or more groups over most of the period and then just Milicia controlling the area. So these maps and clusters turns to maps on neighborhoods. Pretty much we have most of the neighborhoods with at least one group. And we also are able to track the economic, what they are doing in terms of economic activities. So we have a big list and typically like, and we see whether this percentage means that the groups were mentioned that they were exploring this activity. So pretty much what we see is that, as we know, drug gangs are focused on drugs. Drug gangs are focused on drugs and Milicia is on extortion. But we also see Milicia selling drugs and drug gangs doing extortion. And we have a full list of activities they are involved in. And Milicia is much more diversified than drug gangs, but we do have drug gangs investing in several other markets. And we also capture variation over time. There are activities that are decreasing, such as TV internet and transportation, while others such as electricity and water is increasing over time. So what are we doing with this data? So we have a model in the paper that I don't have time to go through it. But basically what we want to model is like the incentives for investing in military capacity or not, what we call fight and the state intervention. So first we do a setup with one criminal group and state intervention and other with two criminal groups and state intervention. So the crucial thing here we want to model is why territorial control matters so much. It's because they derive profits from this territory. There is a cause to fight. And the government in the state, it does have political returns with about fighting these groups. But there is a condition that the groups can also bribe the state in order to not intervene. And there is G, which is our economic losses promoted by the state, which you can think about drugs, seizures, weapons seizures and so on. And what we claim is that the key difference between militia and drug gangs is G, is the fact that because militias are linked to the state, they are able to reduce the economic loss because they got to know when they have or they even can avoid this type of intervention. So what we show is that we have two equilibriums with that can be the violent one and the peaceful one. And for the militia it's easier to get to the peaceful one because not fighting is a dominant strategy even when the state intervene. Then we look at what happened when you have more than two groups and what happens that you have much more incentive to fight. And the reason for that is that you have the constant threat of the rival group to come and conquer your territory. So even though, so it's really what we want to model here is the fact that in this context we have two enemies, the state and the rival criminal group. And this creates a very violent scenario. So with this model we have three predictions. One is that areas with more than one criminal group are more violent because they experience more conflict and state repression. Second is that areas where there is crime consolidation and that we see that is a isolated criminal group. They are less violent if controlled by a militia group rather than a drug gang because they state intervene less and criminal groups especially the militia groups exploit more markets when not facing the threat of a rival group. So how do we test that? We have a panel of neighborhoods. So we regress three different measures of violence on the number of criminal groups and then we split the number of the criminal groups in types of groups. And we have neighborhood and year fix effects. So we are really exploring the change in group composition change in group number to see the effects of violence. So basically the odd columns of this table show one type of regression that has been run by other papers in the literature that basically shows that the increase in the number of the groups are associated with more homicides, shootings and police killings. We use here police killings. It's important to say it's an official measure of police killings. And we interpret that as a proxy of police military repression. And but more interesting in the odd columns we split on we break the, we have indicators for whether the neighborhood has two or more groups, only militia groups or only one drug gang. And the omitted variable here is neighborhoods without any group. So basically we also see as expected that when you have two or more groups you have more violence, but the most interesting thing is the difference of what happened when you have just one group, whether this group is the militia or whether this group is the drug gang. Basically when you have just the militia there is no more homicides and police killings but it does have when you have just one drug gang is associated with 24, 27% more homicides and 47% more police killings. Second, the second exercise we're gonna do is to see how this, the composition of the groups affect economic diversification. While we say it's economic diversification, the type of economic activity they are exploring. So we use this to identify whether the neighborhood indicates that militias or the drug gangs do extortion, drug trade, and then we create a index which are the number of other export markets they do. And here in this exercise the explanatory variable comes from another database from an anthropologist in Rio at the favela level which basically tries to capture this change here. What happened when you are a criminal group isolated from other groups and this and what happened when you have a other criminal group nearby you. And so one is the dummy variable here, it's one whenever there is no other criminal group within one kilometer radius. So what we do is that we do this exercise separated to militia and drug factions. It again has favela fixed effects. Basically it shows that when militias operate without the threat of any other criminal groups around there is increasing mentions to extortion and an additional other market is explored and we don't see this pattern for drug factions. And we interpret that to the fact that even when the drug faction is alone it still has the threat of the state and this is why it's not followed the same pattern. So in order to conclude, so we document that militia groups and drug factions are multi-product enterprises that explore a wide range of listy and listy goods and service. We show that the number of groups is correlated with high level of violence but more importantly we present evidence that neighborhoods with only militia groups have lower levels of homicide shootings but higher level of extortion and economic diversification. And this creates a key challenge when we think about government intervention because this creates a really trade-off that you may have like low violence and higher economic outreach. And a lot of great part of the political sense literature have been defending the strategy of conditional repression based only like focus on group based on low levels in order to reduce violence but here we have a case of a group that is being able to reduce violence but also increasing its economic outreach which is a huge challenge for public safety. Thank you very much.