Why Is Latin America So Violent?

Rights for criminals and limited state capacity are a bad combination

In my forthcoming book, I look for ways to try to steelman the case for populism. One conclusion I come to is that, while populism is usually a terrible idea in developed states, it can make sense in poor nations where things are actually quite broken, and something akin to strongman rule can have beneficial lasting effects.

I think Latin America demonstrates this most clearly. Achieving a monopoly on the use of force is the most basic function of a state. When you see a country with a high murder rate, the costs go way beyond the actual deaths suffered by victims. The numbers are a reflection of the general unpleasantness of life there. Countries with high murder rates suffer from bad behavior that degrades the human experience across the board, from robberies and violent assaults to gang intimidation of normal citizens and loud music in public.

Below is a chart showing GDP per capita plotted against homicide rates across the world. I obtained GDP, homicide, and democracy score numbers from Our World in Data, and dropped countries with fewer than 500,000 people. See here for country codes, which aren’t always intuitive. “ZAF,” for example, is South Africa.

Latin America really stands out. Nations like Brazil (homicide rate: 20.6), Colombia (24.9), and Mexico (24.9) have incomes that are similar to China (0.5), Azerbaijan (2.0), and Georgia (2.0). Honduras (31.4) is about as wealthy as Pakistan (4.3) and Kenya (4.9).

What is going on here? The chart below provides a clue. On the x-axis, it shows the democracy score for each country. On the y-axis is the residual from the chart above. In plain English, the y-axis shows the actual homicide rate minus the predicted homicide rate based on income. Countries above the line are more violent than you would expect based on their living standards.

We see here that, for non-democracies, wealth pretty much predicts murder rate in a straightforward way. Not a single dictatorship has an unusually high murder rate, at least compared to the residuals we get for Latin America. The Substacker Inquisitive Bird has argued that the homicide numbers in Africa are largely made up and can’t be trusted. To test whether this is a problem for my theory, I simply added 10 homicides per 100,000 for each sub-Saharan African country, and the results were similar. The more violent Latin American states still had massive residuals, nearly unchanged.

Even if authoritarian regimes fudge the data, the official statistics roughly fit with what we know about the world. It is assumed by travelers that your risk of being kidnapped or murdered is astronomically higher in Mexico or Colombia than China or Jordan. The numbers here pass the commonsense test.

This article is part of a series discussing topics covered in more detail in my forthcoming book Kakistocracy: Why Populism Ends in Disaster. Please consider preordering here. The release date is July 7, and you don’t get charged until the book is shipped or you receive the digital version. Preorders are very important in determining how much attention a book receives, so by getting yours in you can help ensure that its arguments reach a wide audience.

There are of course many safe democracies, including ones that aren’t rich. Democracy is not a sufficient condition for having an unusually violent country, but it does appear to be a necessary one. We can say that there are for the most part three types of countries in the world:

1. Rich nations, which are safe

2. Authoritarian nations, which are safe

3. Countries that are middle or lower income and democratic, which can be safe or violent

Most Latin American nations fall into category three. The same is true for the three major non-Latin American outliers with unusually high murder rates: Trinidad and Tobago, Jamaica, and South Africa. Each of these gets a Polity score of 9. Note that Cuba, the most authoritarian country in Latin America, is not known for street violence, even though we don’t have numbers. Venezuela, in contrast, has in recent years had an unusual combination of authoritarianism and wide scale violence. But there just aren’t that many dictatorships with a very high murder rate.

Why would democracy be such a good predictor of societal violence? Note that the way we define “democracy” in political science generally includes a respect for civil liberties. This means you don’t torture suspects or keep them locked up without strong evidence. Such nations get warrants before making searches, grant defendants lawyers, and remind them of their right to remain silent.

In a rich country, these are luxuries that society can afford. The police are paid relatively well and get some basic level of training. They have larger budgets with which to fight crime, and, since corruption is more under control, less of it is stolen or wasted. The court system is more reliable, and is pretty good at distinguishing those who pose a danger to the community from those who don’t.

In contrast, if you’re a relatively poor country with few resources and little state capacity, granting protections to suspected and convicted criminals is a major hindrance to keeping order. Law enforcement and court officials can be intimidated or bribed, and their attention and resources are stretched thin. Gangs are able to control more territory, murderers are less likely to be punished, and deterrence breaks down. In recent years in the US, the murder clearance rate has been about 50%-60%. That’s very bad from a first world perspective, but in countries like Mexico and Honduras, it’s closer to 10-20%, and this is quite remarkable given that state officials themselves are often the targets. In Latin America, there are stories of gang bosses running their empires from prison, which would be unthinkable in most dictatorships. The US regularly pressures Mexico to send us their drug lords, because otherwise they might escape from prison at home.

The idea that there’s a tradeoff between civil liberties protecting criminal suspects and public safety is a staple of police dramas like The Shield. The cops pick up some bad character who everyone knows is guilty or has information important for breaking a case. But there’s no proof that could hold up in court. So the antihero turns off the camera, and does what needs to be done. The criminal is like “I exercise my right to a lawyer” and the cop goes “oh I’ve got your lawyer right here” and starts stabbing him with a pen. The case is eventually solved, and we are left to ponder the moral ambiguity of the story. Seems like writers for TV shows are more comfortable with unpleasant tradeoffs than most people who analyze politics.

TV shows are of course fiction, but it sure seems that the world really works like this. For example, before Bukele’s crime crackdown, even after Salvadoran criminals were arrested, they would rely on family members on the outside to keep their gangs functioning and enforce extortion efforts. Now the government has basically picked up everyone who might plausibly be involved in criminal activity, and extortion and murder have plummeted.

So in the case of Bukele, we’ve seen a move away from civil liberties coincide with an increase in public safety. We’ve also observed this process moving in the other direction: democratization accompanied by the breakdown of public order. In 1980, Brazil had a murder rate of 11.7 per 100,000. It inched up the next few years, and in 1985, military rule ended. The homicide rate shot up over the next two decades, reaching 28.9 in 2003. It’s decreased somewhat since then, but the homicide rate remains highly elevated compared to what it was in the last years of authoritarian rule, even though Brazil is 50% wealthier.

Most Eastern European nations remained safe after the collapse of the Soviet bloc. But in Russia, there was a period of lawlessness in the early 1990s, which peaked at a homicide rate of 32 per 100,000 in 1994. Here’s a graph that shows what happened to the Russian homicide rate as democracy increased and then decreased again.

The timing isn’t perfect here, as the numbers went up for the first three years under Putin, but there has been a steady decline ever since.

You’re probably tempted to say that the problem is not democracy and rights for criminals per se, but state capacity. Which is true enough. But, again, it seems like you need the combination of both low state capacity and rights for criminals to get an unusually high murder rate. Authoritarian governments don’t do a lot of things well, but they can at least maintain order, even while being incompetent in other areas. I suspect that this is because maintaining order is easy, relative to all other things governments try to do. When civil liberties protections aren’t in the way, you just need to target the troublemakers, and who they are isn’t an unsolvable mystery. Criminals are pretty dumb, and in some cases, they even get tattoos on their faces and make things as easy as possible for the authorities. El Salvador is an extreme case in the number of mass arrests, but that’s usually not necessary, since under most authoritarian regimes you end up with a less violent equilibrium and law enforcement can zero in on the few individuals who might be a problem. That is to say that Bukele’s roundups probably don’t need to go on indefinitely.

All over the world, populists say that the system is broken, and you need a strongman who is going to bulldoze over opposition and democratic niceties. This is probably not a very good argument in a country where things are going relatively well, like the United States. But if your nation truly is broken, then you may rationally be willing to take a chance on something radical. A country that has a murder rate of 30 per 100,000 should be considered a failed state, and one that is inevitably going to lose legitimacy. There’s a reason that Bukele’s approval rating is almost always found to be over 90%. His popularity indicates that Western elites have massively underestimated the importance of public order.

If a government massacres a hundred citizens, there’s an international outcry. But if it lets an order of magnitude more people be killed in the name of civil liberties, no one cares all that much, and if human rights NGOs express concern, it’s usually over what methods the government might take in response.

Is there something else Latin American countries have gotten out of democracy? Well, according to the World Bank, between 2010 and 2020, Latin America on average grew more slowly than any other region in the world. Such struggles go back decades.

Violence surely must be part of the reason why. Here’s another reason to be open to radical change in Latin America. Governance in that region has failed in terms of not only keeping citizens safe, but also improving their living standards.

I think people struggle with nuance here. Trump, Orban, Bolsonaro, Bukele, and Milei all fall under the umbrella of right-wing populists. Observers tend to have the same view of all of them, either in favor or against. But they each face a set of completely different challenges. Maybe if Trump was born in El Salvador, he’d be a great leader, because you need a thuggish brute in order to get control of a country as violent as that nation was before Bukele. But in America, violence isn’t that big of a problem, we can afford to grant the full menu of civil liberties protections, and the president doesn’t even have much influence over non-federal crimes anyway, so Trump’s thuggish instincts are directed toward trying to lock up his enemies. I probably wouldn’t vote for Bukele if he were running for president of the United States. But someone like him might make a good mayor of Baltimore.

If democracy is a good thing, we may want to unbundle it to exclude soft on crime policies. The way the world works now, leaders who accept a free press and the peaceful transition of power are usually the same ones who don’t adopt authoritarian methods to deal with crime. There’s no natural reason why states that conduct regular, free and fair elections must have high standards for searching criminal suspects and inform them of their right to remain silent after being arrested. Before the Warren Court, the US had much weaker protections for criminal suspects and was less violent, even though it was still a democracy. If global elites insist on the entire package, this can only discredit democracy itself.