October 23, 2002
The Economics of Terrorism
Economist Roger Congleton advances the interesting thesis that we are overestimating and (and by extension, one presumes) overreacting to the risks imposed by terrorism. Sam Sachdeva explains the reasoning in allsci, his interesting webzine devoted to exploring the science behind the news. In a nutshell, the cost in lives from driving and other daily activities far outnumbers the cost of terrorism. The amount of financial and other resources (emotional, psychological) we devote to preventing terrorism is thus disproportionate to the actual risk it imposes.
There is more to Congleton's thesis than this, namely that terrorist organizations are best viewed as political organizations with a purpose, and thereby a certain predictability and rationality which should enable us, if we are willing to set aside our emotional impulses, to deal with them more rationally (i.e., with a clearer sense of costs, benefits, and alternative responses). There is some dissent regarding this view, i.e., that terrorists are pathologically insane and predictable only in that they seek to maximize the horror and death of whomever their twisted religious leaders target.
It is Congleton's economic analysis that interests me. Basically, he suggests that on the issue of combating terrorism we should engage in cost-benefit analysis, which works something like this: one considers the net cost or benefit derived from engaging in a certain activity, after adjusting each for the probability that it will occur. For example, if one goes hunting, there is some probability that one will get shot, and some probability that one will shoot an animal. The economist multiplies the odds of each event by its related cost or benefit, and then makes his decision. This is the reason, by the way, why few economists hunt.
Congleton's approach makes sense, but his analysis is wrong, or at best incomplete. It makes sense because, no matter how emotional we are, we all engage in cost-benefit analysis (albeit poorly at times). Congleton notes, by way of example, that "the risk of hijacking can be reduced essentially to zero by closing all civilian airports." Few people would agree that the risk reduction resulting from such a measure merits its cost, which is exactly Congleton's point -- we are all economists of a sort. Using this methodology, he argues that the risk-adjusted cost (i.e., the cost multiplied by the probability) of driving, for example, is much greater than that of terrorism. He then concludes that the large proposed expenditures to combat the latter are not consistent with our reaction to other dangers (i.e., they are too large by our own safety standards).
But this is only part of the analysis an economist should provide. Recall the second word in the term "cost-benefit analysis." In other words, while one's risk of injury or death from driving is considerably greater than one's risk of a similar outcome resulting from terrorism, driving has large benefits that make its risks more acceptable. In fact, if individual preferences are any guide, then the net outcome of driving is, for the average American, largely positive. Terrorism, however, provides no benefit to outweigh its risk-adjusted cost for the average American.
What's more, Congleton isn't making the proper comparison. If one wants to argue that spending to avert terrorism is disproportionate, one must compare it to the spending to avert the costs of driving (or whatever other activity one wants to use as a benchmark). Basically, Congleton uses in his analysis the risk-adjusted cost of driving without considering the considerable spending to make it safer (beyond highway and road spending, which totaled $112 billion in 2001). In reality, the additional total national spending on airbags, seatbelts, safe car construction, quality tires, driver's education, and the governance of driving behavior must be included. What's more, when he totals the cost of anti-terrorism, he includes a calculation of the opportunity cost created by additional waiting time in airports. Surely we must also then consider the opportunity costs of learning to drive safely, buckling one's seatbelt every day, driving appropriately and at the speed limit, and so on. The point is that his conclusion, that spending to avert terrorism is too high given its risks, is based on inadequate consideration of data.
Even less precise is his consideration of whether the risks from terrorism are increasing. First, he relies on the assumption of the left (though he is no leftist) that economic hardship is a significant predictor of participation in terrorist cells, and then shows that economic wealth is increasing in the Middle East. The available data suggests that his premise is untrue, i.e., the September 11th hijackers and their known leaders benefited from higher than average levels of education and wealth. Worse, he dismisses the belief that technological advances make catastrophic attacks more likely by noting that the hijackers used box cutters and jets rather than an atomic bomb.
The reality is that these catastrophic capabilities exist in increasingly available form, and people who want very much to kill lots of Americans are working very hard to obtain them. It is folly to exclude this from one's calculations simply because it hasn't happened yet. When you factor in the cost of a nuclear detonation in downtown New York City, you have to adjust the risk to darn near zero before measures to prevent it appear too costly.
There's one more flaw in Congleton's reasoning which gets to perhaps the greatest weakness of economic analysis. In a nutshell, economists treat the human mind like a black box. They can't really know who much people value things, so they rely on revealed preferences. Preferences are most readily and repeatedly revealed in economic activity, because it creates a data trail, unlike interpersonal interactions that do not involve recordable exchanges. As a result, the economist is often forced into the box of relying on economic data to talk about things, with a wave of the hand towards the non-economic factors involved.
In Congleton's article this weakness is manifested in the assumption that a death from terrorism carries the same cost as a death from a car accident or even a homicide. I would argue that there is a profound difference in the real cost of such deaths, however. First, we are hard-wired to think in tribal terms. If a fellow citizen whom we don't know dies in a car accident, we think to ourselves, "how sad," and then turn to the next article in the newspaper. When a fellow citizen is murdered by a foreigner precisely because he is a citizen, we take that personally, and want vengeance. (This is worth noting because, while Congleton focuses on homeland security rather than war on terrorism, his conclusion -- that it is too costly -- appears to extend to the latter.)
This leads to the second difference in the two types of deaths, namely, that we are right. I mean by this that our society -- one of liberty and relative tolerance -- is in nearly every respect superior to others. Thus it is appropriate that we treat a death from terrorism as a cost much greater than a death from a car accident, because the former is an assault on our society and way of life. It is evil, in the Biblical sense of that word, and those who engage in it deserve to be killed. Economics provides no method of handling such concepts, and as a consequence economists tend to invalidate them. This is a mistake, because civilizations are built (and destroyed) based on concepts such as these, and an economics incapable of accounting for them is thereby rendered marginal as a tool for social analysis.
That critique aside, I'll bet an economist could show how, in an n-person iterated game (don't ask) that emulates the real world, a vengeance response is optimal. People are less likely to wage war on a citizenry when they are assured this citizenry will wreck extensive and indiscriminate damage on their homeland. I think it's high time we start providing such assurance, and I'm willing to pay, like a lot of other people, a pretty penny to do so.
But then again, perhaps I'm just being irrational.
Posted by Woodlief on October 23, 2002 at 09:07 AM
You're right about the repeated game, and the reason is quite intuitive: establishing a reputation for swift vengeance will greatly diminish the incentive for terrorism. (That's why it's so important to vigorously prosecute terrorists and the countries that support them.) Basically, you're saying that a key flaw in Congleton's article is that it is implicitly a static analysis, which ignores reputation effects in repeated interactions.
As an aside, there is quite a bit of evidence that humans seem to have "vengeance" hard-wired into them. (This may make sense from a evolutionary or social perspective.) By revealed preference, then, humans must derive utility from vengeance ... and it would in fact be possible to derive (from experimental data) imperfect measures of the monetary value associated with this.
Cost-benefit analysis should not be thrown out the window. It's simple rationality: if one behaves in a different fashion, one is being irrational. So the issue is: getting a more accurate measure of all the costs, and a more accurate measure of all the benefits, bearing in mind that some of these are difficult to measure, and bearing in mind that the situation is dynamic rather than static.
Posted by: Randy at October 23, 2002 10:55 AM
While not perfect, the cost-benefit approach does provide a necessary wake-up call. What worries me is that we will likely never be safe from terrorist splinter groups that could (and almost certainly will) come in all flavors.
One problem I have is the widespread belief that terrorist groups are trying to change or destroy our "way of life." While this would undoubtedly be a plus for many of them, I do not think this is their primary motivation or even what they originally set out to accomplish.
What do they want/hope to achieve them?
Rather, it is the simple fact that they gain in stature whenever they attack us. Plus there is the added benefit that attacking us provides them with a steady flow of money, and essentially, they have to attack us in order to keep the money coming in.
Think of the mercenary. He fights for whoever pays the best. While not all the terrorists fit this mold, I'm sure that many do. Case in point, note how the higher-ups never send their children in as suicide bombers--they give that dirty work to the grunts who do it for the glory of Islam.
Obviously, there is some political or theological bent to what they are doing, but there is tremendous stature to be gained by taking on the biggest kid on the block.
Posted by: Trip at October 24, 2002 4:27 PM
Comparing a terrorist death to a traffic accident death, as Congleton does, is not very useful.
It would have been more effective if he had compared a death from terrorism to a death from an alien (foreigner) for economic gain or a crime of passion, etc. In both cases, the victims were innocents. The former case has happens rarely (in statistical terms) while the latter happens far more frequently. Immigration also will arguably have far mor reaching effects on our civilization.
Posted by: Trip at October 24, 2002 5:03 PM
QUOTE: "People are less likely to wage war on a citizenry when they are assured this citizenry will wreck extensive and indiscriminate damage on their homeland."
First of all, terrorist cells and networks such generally do not have homelands which can be ravaged. Even nation-states such as Iraq or North Korea do not really have homelands that concern them -- (at least not in the way Japan and Germany did).
Moreover, the assumption seems to be that punishment will work as a deterrent to future terrorism. However, our penal system would seem to refute such a statement given that our prisons are that are bursting at the seams with people that were not deterred.
Posted by: Trip at October 24, 2002 5:14 PM
Thanks for the comments.
These are among the most thoughtful that I have come across. Two minor points: my original artical in the independent review notes other risks, such as those associated with murder as well. Murder is ethically much more similar to terrorism than car accidents. Second, if you think about worse case scenarious without trying to place some kind of probability on the risks, you'd never get out of bed. The worse case scenario for sending your kids off to school or going to work is limited only by your imagination.
There is also a distinction to be made between a genuine large scale threat in a war that can actually threaten a society, say the level of terrorism in Israel, as opposed to that we have experienced in the US to this point. Obviously, more is effort is justified to counter a large scale threat than to counter a smaller threat. The new budget for anti-terrorism is largley focused on a group that so far has only launched about one significant attack a year is approaching our entire law enforcement budget for all other crimes. We loose six times as many fellow citizens to murder in a typical year as we lost in 9-11, the most deadly year of international terrorism in our records.
Thanks for all the thoughtful comments.
Best regards,
Roger Congleton
Posted by: Roger Congleton at October 25, 2002 10:28 AM
Roger Congleton,
The difference between murder and terrorism is that murder is thousands and thousands of individual, unrelated events. International terrorism has declared us as an enemy - if we ignore that threat, it will grow.
Consider this:
A man owns a house. The pipes under his kitchen sink leak badly - it costs him a hundred dollars a month in water. His water main leaks slightly - a few dollars a month at most in water, but the water leaking undder the foundation of the home. Which leak is more important?
Personally, I would put a bucket under the kitchen sink and have the water main fixed - it's more expensive and saves me little right now, but eventually, it will destroy the foundation of the house, costing me thousands upon thousands of dollars.
You are advocating ignoring the water main leak because there is another leak somewhere else that is bigger - you are ignoring the future.
An imperfect analogy, but I think it gets my point across.
Posted by: Deoxy at October 25, 2002 1:05 PM
1. Cars and murder do not kill thousands or millions of people at a time.
2. Car accidents and murders, no matter how large, do not cause billions of dollars worth of economic damage per event and continue causing such damage at a lower yet still extraordinary rate for (so far) over a year.
3. Neither cars nor even murderers have declared war on America, nor have they been instructed to actively seek to kill Americans, military and civilian, wherever they may be found.
4. Car accidents and murderers do not generally kill in one location, then spread to other locations continuing to kill and maim for days, weeks or months afterward, as will be the case if chemical, biological or nuclear weapons are used.
Posted by: Tom at October 25, 2002 11:12 PM
Deoxy,
The house analogy is nice, but it misses the point entirely. This one is more accurate...
A man has a house. The driveway needs re-paved, the electricity needs re-wiring and the roof leaks. Which is the more pressing?
I don't think anyone is suggesting the terrorism problem be left alone. However, instead of spending billions to invade Iraz followed by the subsequent nation-building, how about spending huge sums to finally get a crack INS department that can manage to keep track of foreigners coming into our country, tighten up the borders, beef-up our intelligence gathering around the world and spend money to "turn" members of terrorist groups while continually harassing them via small-scale, intermittent engagement?
I realize that a "dumb" invastion is the easiest thing to do, but is it the most effective, cost and otherwise?
Posted by: trip at October 26, 2002 12:16 AM
Tom,
Terrorists cannot kill millions or cause billions of damage except when they get extremely lucky which is what happened on 9/11. Eliminating one terrorist group does not eliminate the threat from other terrorist groups.
Also, keep in mind the terrorists are not only after Americans. They have many targets in many countries and generally attack whoever makes a convenient target--Israel and Australia come to mind. The U.S. arguably is more high-profile given its status in the world, but its geographic location still keeps it safer than other "hot" areas.
The nuclear, biological, chemical threat is real. However, it will not go away simply by defeating one terrorist nation or group. The only sound long-term strategy is missile defense, securing our borders, and constant, vigilant intelligence.
Posted by: trip at October 26, 2002 12:50 AM
Trip,
Those things are good, but a secure border can not reasonably defend any port city from a nuclear bomb in a shipping container.
We NEED a crack INS (I'd settle for an existant INS at this point - it would be a large improvement). We need secure borders. We REALLY need good intelligence. Missile defense is a great idea - I hope we get it working very soon.
All of that goes out the window in a nuclear terrorist scenario. We must keep WMDs out of terrorist hands as much as possible. Therefore, we need to eliminate Hussein, who would likely give WMDs to terrorists.
Unless you want to secure every port in the world, a nuclear bomb in a shipping container (plenty of room for radiation shielding in there) can take out an American port city. Come up with a solution to that, and then we'll talk about not taking out Hussein.
Posted by: Deoxy at October 28, 2002 2:21 PM
