Full Site Articles EconLog EconTalk Books Encyclopedia Guides

# Risk of Death for Border Patrol Agents

 Powell's Senate testimony... Hartmann-Sauer ITT...

Alex Nowrasteh has an excellent analysis of the risk of death of border patrol agents, something that has received a lot of attention in the last couple of weeks.

Bottom line: the risk is very low. Alex finds that over the years 2003 to 2017, 33 agents died. That was out of 262,944 for a rate of 0.013%.

A standard way to normalize is to compute deaths per 100,000 people at risk. Doing that, we get 33/2.62944 = 12.55. Round up to 13.

That sounds small. Is it? One way to tell is to compare it to something. In their entry, "Risk and Safety" in The Concise Encyclopedia of Economics, Aaron Wildavsky and Adam Wildavsky have a table showing the annual death rates from various activities and in various occupations per 100,000 people at risk. This compares directly with the 13 number above. The table shows that being a border agent is safer than being a policeman (20) and twice as safe as being a farmer (28). Admittedly, the data in the Wildavsky table are about 15 years out of date and, during that time, due mainly to rising real income (safety is a normal good), death rates in almost all occupations have fallen. But they probably have not fallen a lot. So the bottom line is that being a border agent is not particularly dangerous.

writes:

“That sounds small. Is it?” This is always a good question. It’s nice to have some context for the numbers.

From the Risk and Safety Econlib entry, the mortality for a Chicago crack dealer is higher than for a Texas death row inmate. (7k per 100k vs 5k per 100k). Presumably to deter the crack dealer you’d have to do something to him comparable to the cost he’s currently willing to endure. Which means to get any sizeable deterrence the penalty for dealing would have to be as high as *actually* placing him on Texas death row. Good evidence that supply-side interdiction is probably a losing game, unless the expected legal penalty (*after* multiplying by the probability of capture) is comparable to “a 7% chance of execution.” Wow.

I’ll give my favorite example of death statistics without context. Discussions of the opioid epidemic always list raw totals by year. I’m always left saying, “Is that a lot? What is the risk per user? Is it acceptable?” Some pieces I have read take the dubious step of summing up total overdoses across several years or even decades. It gets you a big, scary number, but completely obscures any concept of “risk per user” of “mortality per 100k population”. The overdose risk per user for prescription opioids looks totally reasonable and is probably worth it for someone who would otherwise experience a lot of pain. Sure, the raw total looks scary until you realize there are ~200 million legal prescriptions and ~80 million legal users in a given year. The fatal overdose risk per user for heroin looks very scary on the other hand, in the 2%-4% per year range; this number is almost certainly driven up by the uncertainties of an illicit market. In most of the pieces I see on this topic, they only give the raw totals without splitting out these very different phenomena or discussing the mortality risk. Raw numbers don't have any policy implications until someone interprets them.

Ralph Truitt writes:

POTUS Obama said there were 21,000 Border Patrol agents yet this flawed study says there are 262,944. I was in Federal law enforcement for 30 years and there are not 262K Border Patrol agents! I understand there are 13K-15K on the Mexican border where most killings take place.

Patrick writes:

[Comment removed pending confirmation of email address. Email the webmaster@econlib.org to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Bob Bell writes:

Ralph Truitt - you raise a logical question. The math is correct but the author left out the explanation for a step. He is taking the death totals for almost 15 years (2003-2017, assuming the totals include 2017 YTD). To adjust to the annual deaths per 100,000 numbers for comparison, you need to either divide the number of deaths by 15 to get deaths per year or sum up the number of agents for the time period to get 262,944 agent-years. Dividing 262,944 agent-years by 15 years yields about 17,500 agents, which seems to be a reasonable number. You have to follow the link to the Nowrasteh article to see the math.

David R Henderson writes:

@Bob Bell,
I was on a flight and just saw this. Thanks for answering Ralph Truitt.
@GregS,
Excellent point. Indeed, I should have written a blog post months ago that says what you said.

TMC writes:

Ralph Truitt has a very good point. I've only known two border patrol agents. One had a pretty cushy boring job, but one patrolled the border. This was maybe 15 years ago. The one that patrolled the border would routinely get into fire fights with Mexican police who would drive into American territory to give cover for drug smugglers.

He was shot once by a sniper, as they have bounties out for them. He was lucky as the bullet went through his windshield and hit his skull it did not penetrate further. The bullet had lost most of its energy from the distance of the shot and the windshield. I know this is an anecdote, but he made it sound like this activity was common.

TMC writes:

In 2011 they had 21,137 Border Patrol agents. (via wikipedia link to cbp.gov

David R Henderson writes:

@TMC,
Ralph Truitt has a very good point.
Ralph Truitt missed the point. See Bob Bell’s answer to him.
On the rest of your comment, as you say, this is an anecdote. For it to be relevant, there would have to be many such anecdotes. As I think George Stigler once said, “the plural of anecdote is data.” We have data. So to challenge my conclusion, you would have to show that the data are wrong. You haven’t done so. You haven’t even attempted to do so.

TMC writes:

Yes, I see what you mean, you are (or the author is) correct. Not sure how I misread that. I did point out my story was just an anecdote just for that reason though. For my friend, firefights, not getting shot, were a normal event.

Another anecdote, so treat it as such, is my cousin. He patrolled the great lakes Canadian border as Coast Guard. He said the main contraband coming over the border was toilets. This was right after the US mandated the low flow toilets here. I guess all border assignments are not equal.

David R Henderson writes:

@TMC,
Yes, I see what you mean, you are (or the author is) correct. Not sure how I misread that. I did point out my story was just an anecdote just for that reason though.
Thanks, TMC. You’re a gracious person.
Another anecdote, so treat it as such, is my cousin. He patrolled the great lakes Canadian border as Coast Guard. He said the main contraband coming over the border was toilets. This was right after the US mandated the low flow toilets here. I guess all border assignments are not equal.
Great story. I bet I will retell it.

Comments for this entry have been closed