El Salvador: the costs and benefits of a crackdown

Out of interest, how much would you pay to avoid a 1 in 68 chance of being murdered?

And how much would you pay to avoid spending a year in prison?

One of the difficult things about being approximately a utilitarian is that you don’t always know whether X is right or wrong. I see a lot of people excited about, essentially, the suspension of the rule of law in order to crack down on criminal gangs in el Salvador. I would like to say that this is a bad thing, but as a utilitarian- while I’m allowed my working hypotheses, I don’t know. I am certainly opposed to what’s going on in el Salvador, but the evidence isn’t enough for me to say that I know it’s wrong. But I think the headline tradeoff that everyone is talking about- the spike in incarceration and the fall in the murder rate is not, of itself, the wonderful exchange that some people seem to think.

My source for much of this is:

https://mattlakeman.org/2024/03/30/notes-on-el-salvador/

Many of those who regard themselves as ‘rationalists’ or even effective altruists support what they see as the successes of the Bukele government with great enthusiasm. My aim is not to relitigate the overall statistics or costs and benefits. I claim no empirical expertise in this area. In this essay I only want to make a really small point using other people’s figures- viz, that the costs of this reduction in murders is not the obviously great deal many people seem to think it is. Imprisoning two percent of the adult population is no small cost.

Background

Before Bukele, under Cerén, murders fell from 107 per 100,000 to 38 per 100,000 between 2015 and 2019. Under Bukele, crime fell in two waves. Up to 2021 it fell to 18 per 100,000. This is roughly in keeping with the trend before Bukele. There is endless debate about to what degree Bukele deserves the credit for this decline, or whether it was only continuing a trend started by Cerén. This is not to say that Bukele isn’t responsible for the decline, but merely to say we can’t know- or at least that figuring it out would take a lot of multiple regressions. The more remarkable phase, the one that’s grabbing everyone’s attention, is the fall between 2021 and 2023:

Over that same time frame, El Salvador’s official murder rate has fallen from 18.17 in 2021 to 2.4 in 2023

This is certainly a break. Consider the fall in proportional terms. Between 2019 and 2021, the murder rate fell by about half, between 2021 and 2023 it fell by more than 6/7ths. Some portion of this fall likely represents the previous trajectory, but for the purpose of this essay we will grant that the decline was entirely caused by Bukele, and in particular, the actions he took in 2021.

How unbearable was the 18 per 100,000 murder rate? It’s pretty high but not the shocking grinder of inescapable death that many people seem to think. There are two states: Mississippi and Louisiana, with a higher murder rate.

How did Bukele do it and what’s the catch? The incarceration numbers jumped by 68,000. Now I don’t want to litigate the rights or wrongs of this in total, but I do want to emphasize that this is not a matter of pushing a button that tells judges to stop being soft on crime or giving the cops a pep talk. It will be important for my argument later that it is certainly not just a matter of finding all the guilty people and giving them a fair trial. If you triple the number of incarcerated people in two years (most of that done in the first year), this is necessarily based on the suspension of the rule of law as commonly understood. A lot of people caught will be innocent, and a lot of the guilty won’t be so guilty as to deserve being locked up in an el Salvadorean prison:

Matt puts it like this:

And those authoritarian measures hurt a lot of innocent people, even in the official government documentation. Of the 77,000 arrests made during the raids, only an estimated 32,331 individuals were full-fledged gang members. Most of the arrests were of “collaborators,” a rather loose term that could refer to a wide range of activities. Many of these collaborators profited, aided, and abetted the gangs, and deserve their fate, but many others likely collaborated under the threat of coercion, or to pay off debts, or were in some other nebulous grey space that exists in gang-dominated impoverished Central American slums.

Many other “collaborators” were unambiguously innocent. Media stories are awash with reports from mothers and fathers whose children were arbitrarily swept up in the raids and have been locked away in hellish prisons for months or years now. Some of these kids just happened to be playing in the wrong place at the wrong time, others were targeted by personal vendettas.

Another factoid that I think captures the sheer absurdity of the situation:

Congress passes bill that could allow 900 people to be tried together if they are accused of being in the same criminal group. If you were the one guy in the hearing who just happened to be in the wrong place at the wrong time and wasn’t a gang member or had the tattoos from a different time in your life that you now regretted, or were coerced into joining, do you think you’d have a fair opportunity for that to come out at trial?

Let’s list some of the costs and benefits of these actions. Sound off in the comments if I’ve left any out:

Broad costs and benefits of the post-2021 Crackdown

Benefits: Approximately 1000 lives saved per year and the averted welfare losses of being dead, the averted welfare losses to families of homicide victims, the averted economic losses caused by the deaths, the reductions in numerous other crimes and their associated costs- probably in total even more significant than murder, greater feelings of safety vis a vis crime, flow-on increases in economic activity, and political stability. The greater cultural harmony, wellness, and other intangibles come from having a low crime rate.

Costs: The financial cost of imprisoning over 1 in 100 el salvadorans additional to the previous rate of incarceration. The welfare costs of imprisoning this many el salvadorans to the imprisoned themselves. The welfare costs for others in particular, for their families. The loss of legitimate employment income that accrued to many of these people and other economic costs. The democratic and civil liberty costs of the creation of a permanent state of emergency. Deaths in prison (greater than 153, but not known), The fear of being swept up yourself because a neighbor says you’re a collaborator, or because you’re in the wrong place at the wrong time. The moral, dare I say spiritual, cost of living in a society where children are imprisoned for having been in the wrong place at the wrong time and where due process has been discarded the name of safety [the irony of people thinking they would walk away from Omelas supporting this! Such costs may seem intangible but I suspect they will eventually return in a more concrete form. Why is it that non-utilitarians always walk away from Omelas the fiction with a strut superiority, but when the city exists in real life, but a worse trade, they run towards it. The child is still a child if they’re in an el Salvadorean prison rather than a basement]

I want to focus in on the small part of the costs and benefits I have some measure of qualifications to assess, partly because of my expertise, and partly because I feel this is what has really grabbed people’s attention viz: THE MURDER RATE GO DOWN, THE INCARCERATION RATE GO UP. People haven’t thought hard enough about these statistics, so lets review them.

The incarceration numbers jumped by 68,000, the murder rate fell by 1000. We can simplify this as follows: for one year of incarcerating 68 people, you can prevent one murder. (Again, we’re just going to take it for granted that this drop was 100% caused by the crackdown). Now maybe this is a good trade. It really might be! Moral uncertainty is the cost of utilitarianism. But what mystifies me is people talking about this tradeoff like it’s manifestly a great deal. Think about putting 68 people in prison to prevent a murder. Now I could tell you that on standard utilitarian methodologies, using QALYs and VSL, this is not going to come out as a positive trade and call it a day. Instead, I’m going to invite you to think through the narrow costs and benefits- ‘just’ of reduced murders and increased incarceration in and of themselves- step by step.

Narrow costs and benefits of increasing incarceration by 68,000 and reducing murder by 1000

First, the benefits. You stop a murder. That’s more life for someone, murder victims are often young, so let’s say 45 years on average. You save the grief of their family and friends. Finally, there is the economic contribution of the person you have saved.

Now the costs. 68 people spend the year in prison. How much would you pay to avoid spending a year in prison? Would you prefer a 1 in 68 chance of dying or a 100% chance of spending a year in prison? I know which I’d take. Now think about the economic costs of having to pay to keep people locked up for a year. Now add in the reduction in economic activity as people can no longer work the jobs they worked before they were imprisoned. Think about the reduction in the welfare of their families- both through familial bonds and poverty.

Perhaps you think that the welfare of these prisoners does not matter because they have done bad things, in which case I would remind you that there hasn’t exactly been due process here. If you get to put your thumb on the scale by insisting welfare costs to the guilty don’t matter, I get to put my thumbs on the scale by saying punishing an innocent person, especially without proper process, is many many times worse than a misfortune happening to an innocent person organically. Also, if we’re going down this road of devaluing the welfare of criminals, which I don’t support, we should remember that the murder victims are strongly disproportionately gangsters themselves.

The headline exchange that’s got everyone talking- 68,000 extra prisoners at any given time of year for 1000 fewer deaths per year - is, prima facie, not great. If you put a gun to my head I’d guess it’s off by a factor of five or so. There are numerous other factors here, from every other crime apart from murder to the political costs of an authoritarian government in the long term, but I’ll leave that for others.

Henry comments

I asked Henry, my friend the bartender at the bar I go to to not drink alcohol at and hang out (seriously, I do this almost every day, I don’t know why), and I asked him what ratio he’d accept of prisoners to murders prevented. Henry’s response was that, fundamentally, it is not healthy to think in terms of such a ratio. You simply do not want to go down that road politically or morally.

I think there’s a wisdom in this. I’m not rejecting the idea of a cost-benefit analysis of prisons, I support that. I’m not even necessarily against thinking in terms of such ratios as a guide to policy to some degree. But my gut says nothing good comes out of dreaming about a higher incarceration rate and making that itself a target. Fundamentally, it is a lawless, loveless way of seeing things, and in the sweep of our history, law, and love are both hardwon achievements, more fragile than recognized and sometimes most at threat from their preachers. Although I cannot know, I am more afraid of what lies beyond their boundaries than within them.

.