Bryan Caplan  

Terrorism and Innumeracy

PRINT
Antitrust vs. "Cannibal Capita... Smith's Defense of Western Civ...
Maverick Philosopher Bill Vallicella responds to my praise of Nathan Smith on terrorism:
The Caplan/Smith argument is that because the number of auto-related deaths is much greater than terror-related deaths so far, a high level  of concern about terrorism is not objectively warranted.

But this sort of reasoning involves  vicious abstraction.  It is highly unreasonable to consider merely the numbers on both sides while abstracting from the motives of the terrorists and the societal impact of terrorism.    With very few exceptions, drivers do not intend to kill anyone, and when their actions bring about deaths, those deaths involve only themselves and a few others.
I happily admit that Nathan's case would have been stronger if he'd focused on garden-variety wrongful killings (murder and non-negligent manslaughter) rather than accidental killings.  But the substance stands: Despite the worst terrorist attack in history in 2001, terrorism sums to just 1.5% of U.S. wrongful killings in the 21st century.

Vallicella:
One cannot reasonably abstract from the political agenda of terrorists and the effects even a few terrorist events have on an entire society.  Ask yourself: has your life changed at all since 9/11?  It most certainly has if you travel by air whether domestically or internationally.  And even if you don't.  Terrorists don't have to kill large numbers to attain their political goal and wreak large-scale disruption.  The Tsarnaev attack on the Boston Marathon shut down the city for a few days.  Same with Paris, San Bernardino, Madrid, London, etc.   That had all sorts of repercussions economic and psychological.  

And if you care about civil liberties, then you should take the terror threat seriously and do your bit to combat it. For the more terror, the more government surveillance and the more infringement of civil liberties.

This boils down to complaining about the reaction to terrorism.  But that's ultimately my and Nathan's point: Popular cures for terrorism are far worse than the disease.  "Terrorism is a terrible problem we must fight with everything in our power" and "Terrorism is a terrible problem because we want to fight it with everything in our power" sound alike, but they're diametrically opposed positions.

Vallicella's strongest point:

There is also the obvious point that jihadis would kill millions if they could.  Would they use nukes against the West if they could? Of course they would. And that would change the raw numbers!
I think the same goes for many, if not most, non-terrorist mass killers.  Anyone who murders a classroom of children for fun might detonate a nuke for fun.  That too would change the raw numbers!  I will admit that the odds of a terrorist obtaining a nuke seem much higher than a mere sociopath doing so, but both are extremely remote.




COMMENTS (24 to date)
matt H writes:

The chance of a non-state actor getting their hands on a nuke, seems closer to one-in-four to me.

What odds would you give of Pakistan getting taken over by radicals, who give the bombs to ISIS. Do you really think this is a remote possibility? What about the Saudi's building nukes, what chance does that regime stand of lasting another 20 years?

What about these groups just creating a dirty bomb using medical waste.

I don't think we should over-react to the threat of terrorism. The world has had waves of it before. 100 years ago it was anarchists bombing subways and market squares. Islamic terrorism will subside too. But if these threats aren't keeping you up at night, you haven't thought them through enough.

Maybe you should make a graph of possible outcomes for ISIS land holdings in 20 years, assign probabilities to each. Here's one ISIS becomes what the soviets where in the 20s, 30s and 40s, monstrous, but with lots of support outside their borders. With the addition of a nuclear weapon it liberated from a failed state, and supporting terrorism globally, but protected at home by a nuke, give odds on that. I'd say 1 in 10. What would we do if North Korea inspired terrorism around the world?

Thomas writes:

The point of comparing terrorism-caused deaths with, say, deaths in automobile accidents eludes me. They aren't mutually exclusive. Terrorism simply increases the possibility of being killed. Whether the cost of trying to prevent death by terrorism is justified by the effectiveness of anti-terrorism measures is another question entirely. And its a question that only those who bear the costs (and risks) are qualified to answer.

Ben Kennedy writes:

There are concerns beyond raw numbers and probabilities. The (supposed) endgame of terrorism is to destablize society. The endgame of a mass shooter seems to often be suicide, which is not really threatening the fabric of society. Of course there is no endgame with automobile deaths, it is just bad stuff that happens. Fear of terrorism is not simply fear of getting personally blown up, but fear of systematic attacks on institutions. Now, one can argue that this fear is overblown. My point is that story is more complicated than "human beings are bad at math"

Mark Bahner writes:
The chance of a non-state actor getting their hands on a nuke, seems closer to one-in-four to me.

Yes, I'd say that the chances of a non-state actor exploding a nuclear device in this century are probably greater than one-in-four. But even if they killed 20,000 people with one, that's about equal to a couple years of drunk driving deaths in the U.S. (and less than a year's drunk driving deaths world-wide).

What about these groups just creating a dirty bomb using medical waste.

This is not a subject on which I have expertise, but as far as I know, "dirty" bombs have little chance to kill more people than "clean" bombs of similar explosive power.

So let's say there is a 500 pound bomb that's "dirty." The deaths caused by that are not going to be even close to a very small nuclear bomb, which would have 1000 tons or more of explosive power.

But if these threats aren't keeping you up at night, you haven't thought them through enough.

The danger of my dying in a car crash isn't keeping me awake at night, and that's far, far more likely than my dying from a terrorist nuclear device or "dirty" bomb.

"Maybe you should make a graph of possible outcomes for ISIS land holdings in 20 years, assign probabilities to each."

According to Huffington Post in August 2015, ISIS controlled somewhere between 13,000 and 35,000 square miles. Let's call it 20,000 square miles.

I predict that in 2 years, the chances are 50/50 it will be ISIS will control less than 10,000 square miles, an in 20 years, the chances are 50/50 they will control less than 1000 square miles, and a 90 percent chance it will be less than 10,000 square miles.

Here's one ISIS becomes what the soviets where in the 20s, 30s and 40s, monstrous, but with lots of support outside their borders.

That doesn't seem plausible to me. The Soviets controlled far more territory and had far more sympathizers outside their borders than I can conceive of for ISIS.

Mark Bahner writes:
The point of comparing terrorism-caused deaths with, say, deaths in automobile accidents eludes me.

The point is that economics is all about using scare resources (e.g., time and money) to address a wide range of problems.

It doesn't make sense economically to spend a lot of time and money on something that doesn't kill or injure many people, with the result of not having time and money to spend on things that do kill and injure many people.

pyroseed13 writes:

Bryan, the problem with your argument is that you assume terrorists deaths are exogenous with respect to our institutional arrangements, but they clearly are not. How do we know that number of terrorist-related incidents and deaths would not be higher if we stopped doing everything we currently do to prevent terrorist attacks? As Vallicella's expressed, if terrorists could kill even more than they do they probably would.

Ben Kennedy writes:

"It doesn't make sense economically to spend a lot of time and money on something that doesn't kill or injure many people, with the result of not having time and money to spend on things that do kill and injure many people."

This reminds me of Scott Sumner's recent point on uncreative economists. Playing the lottery may be irrational if considering only dollar figures and probabilities, but it is rational when considering other hidden utilities. Fear of terrorism may be irrational when considering only lives lost, but there my be hidden disutilities to take under consideration. I think the interesting question is, "why does is the disutility from terrorism-related deaths so much higher than car accidents"

Tu Fu writes:

Ross Douthat already explained why terrorism is different than other deaths:
http://douthat.blogs.nytimes.com/2015/11/30/what-we-fear-when-we-fear-terrorism/?_r=0

Jeff writes:
I happily admit that Nathan's case would have been stronger if he'd focused on garden-variety wrongful killings (murder and non-negligent manslaughter) rather than accidental killings. But the substance stands: Despite the worst terrorist attack in history in 2001, terrorism sums to just 1.5% of U.S. wrongful killings in the 21st century.

I think what you, Bill, and Nathan are all overlooking is the following: under a policy of open borders, the number of car crashes in the US would probably rise, as highways and streets get more crowded. That said, people can compensate by driving slower, swapping Porsches for minivans, taking public transit more often, travelling at off peak hours, etc. The result is that the absolute number of deaths increases, but the rate of deaths per 100k vehicle miles or whatever metric you wish to use probably wouldn't change that much.

I don't see how you can really make a similar claim about deaths from terrorism. A western country with open borders is going to see a large influx of Muslim immigrants, some fraction of whom (or their descendants) are going to decide to engage in acts of terrorism. As the Muslim fraction of the population grows, the frequency of terrorist acts likely grows, too, and the steps necessary to avoid this happening have really high personal costs like erosions of civil liberties, TSA checkpoint proliferation, etc.

Ahmed writes:

"Secondary causes are only a veil to occupy the common people. God's elect see through the causes, to the Causer of causes."

—Rumi

Islam is an occasionalist religion, i.e., everything comes from God. And one of God's names is "The Giver of Death". And he has no partner. Death comes from God and God alone. In Islamic theology, the idea of a life being cut short is nonsensical. Each of our lives is fixed by Divine decree. When our time comes, we are taken to our place of death.

To be afraid of terrorists, or anything else for that matter, is ignorance. For it is to be afraid of that which can not harm you. It confuses the agent with the act. Which proceeds from God.

Hence the saying, he who does not fear God will end up fearing everything else. Like terrorists and car crashes.

The occasionalist way of thinking about God is the only true one. If you think otherwise, you will end up with the problem of theodicy (i.e., how can a loving God allow evil). And that will lead you to one of two conclusions about God:

1) God is benevolent but impotent.

2) God is omnipotent but malevolent.

The first God is not worth praying to. He couldn't help you even if He wanted to. The second God, well, he might smite you just for having the nerve to ask for help.

We believe in a God who is both benevolent and omnipotent, i.e., the occasionalist view. And when we look around, we see nothing but the face of God.

Charley Hooper writes:
Despite the worst terrorist attack in history in 2001...

I'm no expert on the subject, but I don't think this is even close to being true. There is always a definitional problem with terrorism, but I think it is an old and widely used technique. Here is just one example.

In 88 B.C., Mithridates VI Eupator Dionysus, King of Pontus, took advantage of Roman problems at home by sweeping through the Roman province of Asia Minor. So swift and successful was his conquest that many thousands of Roman citizens and their Italian allies were unable to escape. Mithridates solved the problem expeditiously. He ordered that every one of them should die at an appointed hour on a single day, throughout the province. An estimated eighty thousand Romans perished that day.
Charley Hooper writes:

There are two main reasons I'm not very concerned about terrorism:

(1) Terrorists could kill a large number of people if they did X, Y, and Z. But they've been trying to do these things for decades and haven't been able to.

(2) Compared to the Soviet Union during the Cold War, terrorists are impotent amateurs. The Soviets disliked us and had the ability to obliterate us, either on purpose or by accident, and yet we are still here.

Conscience of a Citizen writes:
This boils down to complaining about the reaction to terrorism. But that's ultimately my and Nathan's point: Popular cures for terrorism are far worse than the disease.

The popular (partial) cure for terrorism would be to exclude likely terrorists from the US. But you (Caplan/Smith) utterly reject that. Instead you advocate importing likely terrorists, then saddling the US with the TSA, the ($40 billion/year) Department of Homeland Security, mass surveillance, hundreds of thousands of tax-eating armed guards standing around DMV offices, etc.

Suppose that just half of the DHS budget is justified by "combating terrorism." That's $20 billion. That is also 25% more than the total economic value of the estimated two million immigrants from majority-muslim countries currently in the United States!

Check the math yourself (I have to use all sorts of averages here, but the law of large numbers says that's okay). Of the two million immigrants, 800K do not work (Census). The workers support the non-workers so we'll average in the costs of social spending. The 1.2 million workers earn around $30K each (Census), so generate surplus for their employers of ~$20K each (60% labor share of income). But they cost taxpayers ~$7K each in subsidies after deducting the taxes they pay (several sources). So the net surplus is ~$13K each. Total net surplus (from the two million immigrants) is around $16 billion/year. Of course, most of the surplus accrues to employers, but neglecting distributional questions, it is obvious that the muslim-country immigrants do not even pay for a fraction of the costs generated by their propensity for terrorism.

Employers pay around 25% of their surplus in taxes, say $4 billion, which means that median Americans have to pay another $4 billion out of their own pockets just to cover the ~$8 billion cost of social spending on the two million immigrants.

To restate all of that, muslim-country immigrants have significant negative value to most Americans. Worse, to help muslim immigrants provide $12 billion surplus (after taxes) annually to American plutocrats, non-plutocrat Americans must pay $4 billion, about the most regressive transfer scheme you could ask for!

Emily writes:

I'll take the car analogy seriously. The total number of fatalities from terrorism vs. automobiles should not be highly relevant to anyone's decision-making: we're should be interested in the marginal costs/benefits of various actions.

We have already picked the low-hanging fruit when it comes to auto safety: we already spend an enormous amount to reduce fatalities from car crashes. It's a highly regulated area, and people often go beyond the mandates and spend more money for a safer experience. And the benefits from auto travel are both enormous and largely reaped by the people who are also bearing the costs in terms of possibility of death.

The low-hanging fruit when it comes to at least not increasing the expected number of terrorism deaths is not increasing Muslim immigration to this country. This isn't about the total deaths from either, it's about the marginal action (another immigrant), the marginal cost (risk of terrorism) and the marginal benefit (largely accruing to the immigrant, not the people who are increasing their risk of dying through terrorism.)

Another crucial distinction between terrorism deaths and other wrongful deaths is, of course, the random nature of the former. Murder in this country is highly concentrated, and you can do a lot to avoid it by who you associate with and where you live. (And people do! They spend enormous amounts to live in safer neighborhoods.) But unless you are going to avoid public spaces, you cannot reduce your probability of being a victim of terrorism.

Mark V Anderson writes:
Fear of terrorism may be irrational when considering only lives lost, but there my be hidden disutilities to take under consideration. I think the interesting question is, "why does is the disutility from terrorism-related deaths so much higher than car accidents"

Yes, Ben, this is exactly the question. Why do so many find more disutility from terrorism deaths than auto deaths? (Or individual crime deaths?) To me this makes no sense at all. I thought you were one of those who does find more such disutility, so I would find it very useful if you would answer your own question.

In an earlier posting you mention the problem of destabilizing society, but again that is because of the reaction to the terrorist deaths, not the deaths themselves. If we treated terrorist deaths with the same resources we put to other equivalent levels of deaths, there would be very little destabilization. Again, we are back to the irrationality of treating such deaths differently.

Mark Bahner writes:
But unless you are going to avoid public spaces, you cannot reduce your probability of being a victim of terrorism.

The probability is essentially zero, so why would anyone try to reduce it?

Ethan writes:

The reaction is indeed the thing to look at. If you were to deal with terrorist acts, not as terrorist acts, but criminal acts, then it is no longer terrorism.

Lock the individuals up, take them out, whatever needs to be done. Enact justice properly without apology, and without straying from basic natural principles and move on. If we had a fair justice system, capable of taking care of security, not dependent on the fear of the people, terrorism as a term and concept would essentially not exist.

Nathan W writes:

Terrorism sells newspapers and certain political groups abuse these events for political gain.

Fear-based politics seem to work. It would be better if history proved that people who use fear for political gain fail, not succeed.

Ben Kennedy writes:
Yes, Ben, this is exactly the question. Why do so many find more disutility from terrorism deaths than auto deaths? (Or individual crime deaths?) To me this makes no sense at all. I thought you were one of those who does find more such disutility, so I would find it very useful if you would answer your own question.

I personally don't fear terrorism from a personal or societal standpoint. I agree that domestic terrorism reactions are in some way destabilizing. What I'm talking about though is the supposed clash of civilizations, in line with Arnold Kling's civilization/barbarism axis of politics. Basically, the fear that ISIS or some similar radical group sets up sharia law in Delaware. You and I think it's silly because we don't view the world that way, but that is the main fear narrative - not personal death at the hands of ISIS. Through the civ/barb axis, the potential disutility of terrorism is enormous.

If I really wanted to comfort someone who was afraid of terrorism, my tactic would not be "cheer up my innumerate friend, you're more likely to die on the way to work than be killed by ISIS". I would probably say, "Oh, ISIS can't really take over here, they don't even have boats. And if things really get hairy, we can always nuke them"

Conscience of a Citizen writes:

I made an arithmetic error above, which is particularly embarrassing since I introduced it with "check the math yourself!"

My error affected a couple of numbers but did not lead to an incorrect conclusion. Correcting my error actually shows immigration to be a worse deal for median-Americans.

Here is the error: while calculating the surplus from immigrant labor I subtracted the public subsidies to the immigrants from the nominal surplus. That was okay for figuring the total net surplus, but when I subsequently tried to assign that surplus between plutocrat- and median-Americans I forgot to add the public subsidies back in before calculating plutocrats' taxes and profits.

Here is the proper calculation (all numbers annual): of 2 million muslim immigrants, the 1.2 million who work earn ~$30K each, so generate surplus for their employers of ~$20K each (60% labor share of income). Total employer surplus is ~$24 billion. Employers pay taxes of about $6 billion on that amount, leaving them with $18 billion. But the immigrants consume about $8 billion in public subsidies (above what the immigrants themselves pay in taxes). That forces non-employers to pay $2 billion in extra taxes to support the 2 million muslim immigrants.

The total net surplus (~$24 billion minus ~$8 billion) is still ~$16 billion as I calculated previously, which is less than even a portion of the cost of muslim terrorism.

But worse, the correct calculation shows that the $2 billion which median Americans pay in extra taxes goes directly to immigrants' employers! The employers end up with $18 billion (after taxes, remember) even though the total net surplus from the immigrants' labor is only $16 billion!

No wonder employers are so eager for immigration. Under the current American constellation of taxes and welfare schemes, every single average immigrant brings employers $1000 after taxes at the expense of median-American taxpayers!

Emily writes:

Well, Mark, I think we bear various costs to reduce the probability of already very-unlikely bad things. For instance, even a very unsafe neighborhood, your probability of being murdered is extremely low if you are not involved in gangs or drug crime, do not associate with people involved in those things, and do not resemble people who are. But on the margins, people still incur costs to lower than probability even further by avoiding those neighborhoods.

Whether you should incur that type of cost depends not just on how much you can reduce the probability (what the marginal benefit it) but also on the marginal cost of that action is. I wouldn't suggest at this point that anyone be performing personal actions to reduce their likelihood of being a victim of terrorism while in the United States. I am suggesting that when we talk about things that would increase the expected number of terrorist victims, that increase should certainly be part of our analysis, even if terrorism would remain low in the scheme of things-people-die-of.

Hugh writes:

Figures from past years are probably a good guide to forecasting car fatalities: there is a huge population of cars and each accident is a discrete event.

But can we say the same thing with terrorism? Is the past a good guide to the future? I say no. If, in the year 1943, you had asked an American to rank Germans, Japanese and Muslims in terms of risk to himself and his country, the Muslims would have come last - just look at the (then) recent statistics!

Fast forward 70 years and the 1943 statistics are no longer relevant as times have changed.

Americans are right to be worried about terrorism: the past confers no comfort on the future.

Vasilis Kostelidis writes:

Please forgive me for my bad English.

This is outdated, but I just finished reading the post and comments.

In my opinion, it is not the work of an anti-terrorist government that prevents terrorism.
It is the knowledge that anti-terrorist measures exist, that mainly reduces terrorism. It makes sense. Many potential terrorists will not try to enter a country with tight border anti-terrorist control like the USA, because they know that their plans will not succeed.

I think I didn't read this argument in any comment above, but it has been mentioned by someone, I apologize.

Adam writes:

Bryan wrote:

But the substance stands: Despite the worst terrorist attack in history in 2001, terrorism sums to just 1.5% of U.S. wrongful killings in the 21st century.
As another commenter pointed out, there's no requirement that society as a whole losses the same quantity of utility for every death. There is an established concept of "dread risk" which is basically that the utility-loss-increase with number of deaths is not linear. People care more about multiple-fatality accidents (e.g. plane crashes) than just adding up the equivalent number of individual, accidental deaths. So attitudes to terrorism are not unusual in this respect and people aren't necessarily being innumerate. They're just attaching different values to different kinds of deaths.

Adam

Comments for this entry have been closed
Return to top