Garett Jones  

Will Profiling Even Work? A Supply-Side Perspective

PRINT
Carnegie on the School Ethic... My Election Outcomes: Final Up...
Bryan takes me to task for my prediction that free market airline security would still be rigorous and intrusive.  His prediction: 

..here's one massive cost-cutting, convenience-raising change I'd predict in a free market: Profiling.  Private security firms would still claim to treat everyone equally.  But they'd wave the elderly, women, families with children, and well-dressed men right on through, especially if they look like "regular Americans."  

Of course, actual terrorists have already figured out the well-dressed thing; here's one person going through security on a certain September morning we all recall; these two people boarded AA 77 at an airport a few miles from my house the same day.  All three quite well dressed. 

Here's another hypothesis put forward by Bryan:

Private security firms would redirect their spare attention to (a) young men who look like they might be violent criminals, and (b) anyone who looks like they might be of Middle Eastern descent.

Let's focus on case (b).  As a matter of crude efficiency--leaving aside the salient issue of fairness as far outside my scope of expertise--would rigorous profiling actually work?

It depends partly on how readily al Qaeda and similar organizations can find substitutes for men who fit the profile.  You might think that we can look around at recent attempted terrorist attacks--the Shoe Bomber and the Underwear Bomber--and conclude that Bryan's forecast is already a decade out of date. 

But things aren't that bad for Bryan's hypothesis: They're worse!  And basic price theory helps explain why. 

Let's consider two chalkboard regimes: Under one, the government treats every passenger identically, and under the other, the government focuses all attention on passengers fitting an ethnic profile.  Clearly they're both extreme cases, and actual U.S. experience and potential profiling regimes are both somewhere along that continuum.  

If a government credibly commits to treating everyone identically, what incentive does a terrorist organization have to search for recruits who differ from the stereotypical terrorist profile?  None whatsoever.  All passengers will be treated identically, so if it's even slightly easier to find recruits who are part of the profile (a big if), then 100% of actual airline hijackers and bombers will fit that profile.  Lowest cost sourcing.  

However, if the government (or a private security firm) switched to 100% profiling, then it's obvious what the terrorist organization will do: Find someone outside the profile.  Past experience suggests al Qaeda has already done this, and just last week a bit more evidence came in supporting the hypothesis that ethnic profiling at the airport won't be quite as effective as some people think, even under our current rules.  And if ethnic profiling became legal, terrorist organizations would have an even stronger incentive to recruit from non-profiled demographic groups.   

Hijacker ethnicity is an endogenous variable.  

Coda: Info on female suicide bombers here

Comments and Sharing





COMMENTS (40 to date)
John Thacker writes:

A fine and persuasive argument. However, you haven't that I have seen addressed the argument that private security (not just government contractors) would be less intrusive because it would respond to the desires of customers instead of voters.

William Barghest writes:

This post is bizarre.

I'd say the relevant question is whether profiling decreases the security costs of the airport while increasing the recruitment costs of the terrorists. You do not argue against this but merely state that since terrorists who do not fit the profile "can" be recruited, this means that profiling does not work.

Tom West writes:

Well, if you believe the whole thing is "security theater", then profiling works for private enterprises because it targets the people who the "average Joe" customer thinks looks like a terrorist, making them feel safer.

This also means that "average Joe" is not personally inconvenienced, yet it allows him to feel that he's secure.

Of course, whether profiling (or any other security measures, for that matter) *are* safer is irrelevant to customers and dependent on so many variables, we can never get a straight answer anyway.

MikeDC writes:

Customers of airline flights already have a much stronger desire to see that the plane arrives safely than the average voter.

Finch writes:

> Customers of airline flights already have a much
> stronger desire to see that the plane arrives
> safely than the average voter.

That's not obvious, since the costs of a successful terrorist attack on an American commercial aircraft would likely include a war, the effects would be spread very widely and the death toll would likely significantly exceed the number of people actually on the plane. You might even want to include the effects on whomever it is we would be bombing.

Airline customers are presumably richer than average and might be slightly less injured by these other costs of a downed airliner, assuming as is likely that they aren't on that specific downed airliner.

Gordon writes:

The argument for profiling operates under the assumption that only willing recruits are the ones who will be used to take out the plane. You could have attempts to use unwitting or unwilling recruits. I vaguely recall a case in Israel in which a terrorist operative had romanced a woman who did not fit the profile in order to secrete a bomb in her luggage. Or terrorists could simply threaten people's loved ones in order to induce them to carry explosives onto an aircraft.

The use of unwitting or unwilling recruits seems like a very low cost alternative to deal with a security system that uses profiling.

MikeDC writes:

Finch,
No, I wouldn't include the effects on whomever it is we would be bombing since the comparison was between flying customers and voters.

assuming as is likely that they aren't on that specific downed airliner.

Unfortunately, this exactly the the point at which it really does become obvious that customers of airline flights have a much stronger desire to see that the plane arrives safely than the average voter.

Suppose you get to the airport and your airline announces "Hi passengers. Would you like your flight to receive a security screen today?" Is it conceivable that customers of a given flight will ever collectively buy the "no" option?

------------------

Profiling, like any other strategy, can only be worthwhile to a point. Once the opponent adopts to the strategy of profiling, your costs go up because you can no longer rely just profiling.

Arthur_500 writes:

You have a fair argument for one specific piece of the security puzzle. the problem is that security is mostly a joke. It matters not who is doing the security what matters is the incentive to do well.

Let's say that the Security firm were capable of being sued if a terrorist got through.

Now you will get real security.

The security firm will utilize a matrix of procedures to assure itself that they are getting through to terrorists in an efficient manner. It not only saves money but streamlines actually getting secure results.

Customers will complain but then they are the first ones who will file suit if the security measures fail.

dave smith writes:

Question for Garret: IF people that fit the profile were indeed "low cost terrorists," then wouldn't there be reason to prefer the private security equilibrium since terrorism would cost more?

David Johnson writes:

What is the goal of the security? If it is ONLY to catch a certain brand of sophisticated Islamist terrorist, then profiling will not be sufficient to catch them. But if the goal is also to reduce criminals in general from boarding the planes, then profiling is indeed a useful tool.

Ken B writes:

Garett is assuming a pretty crude form of profiling isn't he? After all the airline has your ID and can check you in many ways. They can even make searchability a requirement to fly if it proves difficult. It's hard to hide your secrets from a good quant.
I suspect a good automated search can infer a high level of religiosity and adherence to a given religion.
Other indicators of conversion might include divorce, etc.
Like it or not, that's what the profiling is for.

David Friedman writes:

At a slight tangent ...

The clearest evidence to me that TSA does not care about the welfare of passengers is their failure to take the most obvious precaution against pilferage or vandalism of searched luggage. That precaution is for the note left in searched luggage to identify the agent who searched it, something the private firm that does security at SFO already does, and something lots of other private firms in related contexts routinely do. It costs almost nothing and makes stealing from luggage considerably riskier.

My guess is that it is not a mistake, that the organization would rather not be able to identify agents who engage in such behavior, since if they could they might have to do something about it.

That suggests ways in which privatizing would improve things—reduce costs to passengers whether in pilferage or hassle—independent of whether it improved security.

Justin writes:

If profiling at the airport actually lead terrorist groups to seek out different demographic groups, that would be a significant boon to law enforcement. Organized crime groups (terrorist or otherwise) that are based on an insular ethnic group are generally much harder for law enforcement to infiltrate. If terrorist groups started recruiting average-looking Americans, it would be much easier get a few FBI (and other) agents under cover where they could do some real good by uncovering plots and gathering intelligence. As long as terrorist groups only need to recruit from populations that are difficult for law enforcement to infiltrate because they're based on long-standing family ties, they have a significant advantage.

Of course, personally, I tend to believe that passenger screening adds essentially nothing to actual security-- anyone smart enough to evade the police while plotting an attack isn't going to get picked out by any plausible airport security regime though there is a good chance that they'll be stopped by passengers in flight. So to the extent that profiling leads to most people spending less time in security lines, it would work in economic efficiency terms.

Silas Barta writes:

Am I missing something here? Forcing terrorist organizations to find someone who "doesn't look like them" would raise their costs of launching an attack -- and, for that matter, make them easier to infiltrate, as they have to be "more open to diversity!"

Is this not a positive outcome from a security perspective? It's a lot easier to launch an attack if you can just use a committed, regular member of your group, than if you have to recruit Grandma Frail to sneak your things past security.

Justin Talbot writes:

@Silas

Yes, profiling raises the cost of an attack by making the terrorists find someone who "doesn't look like them". But profiling also lowers the cost of an attack. Profiling assumes that people who don't look like terrorists will receive less scrutiny than they do now. Thus, attacks by such people will be more likely to be successful than they are now (or under a purely random scheme).

The question is whether the combination of these two effects increases or decreases the *net* cost of terrorism.

ColoComment writes:

The general assumption seems to be that wannabe terrorists will again target an airplane in flight.

If I were a terrorist intending to kill a lot of people, scare the bejeesus out of the flying public, and generate chaos in the U.S. airport/airline industry, I would avoid the security lines altogether.

I could blow myself up while WAITING in a crowded security line, while standing near the most crowded of the baggage claim stations, wile sitting in a pre-security restaurant, or the like.

My intent would be to leave the flying public wondering not if their airplane is safe, but if anywhere in the airport is safe.

How're ya' gonna defense that?

Joe Cushing writes:

What makes flying safe now is that passengers are no longer following the advice of government to be passive in the face of threat. Passengers are fighting back. I'd like to see concealed carry legal on planes.

ajb writes:

And yet the Israelis seem to use profiling to good effect. Why not us? Even if it's just no better than the current system, if it costs less, isn't that a win for us?

Steve Sailer writes:

Indeed, Mohammed Atta set off the terrorist intuitions of the desk clerk who checked him, but who refused to act on it out of political correctness. From Oprah.com in 2005;

"Michael Tuohey was going to work like he had for 37 years, but little did he know that this day would change his life forever. On September 11, 2001, Tuohey, a ticket agent for U.S. Airways, checked in terrorist Mohammed Atta for a flight that started a chain of events that would change history.

"Tuohey was working the U.S. Airways first-class check-in desk when two men, Atta and his companion Abdul Azziz-Alomari, approached his counter. From all outward appearances, the men seemed to be normal businessmen, but Tuohey felt something was wrong.

"I got an instant chill when I looked at [Atta]. I got this grip in my stomach and then, of course, I gave myself a political correct slap ..."

http://isteve.blogspot.com/2006/09/political-correctness-saved-mohammed.html

Jeff writes:

I think Justin Talbot pretty well nailed it.

Tom West writes:

I'd like to see concealed carry legal on planes.

How many planes do you think it would down when out-of-their-mind drunks put a round through the wrong place in the plane?

Thanks, but I figure it would beat the terrorists total of destroyed planes by an order of magnitude, given there's a few emergency landings each year necessitated by essentially insane drunks.

Steve Sailer writes:

Dr. Jones' argument isn't very good. It is highly costly for al-Qaeda to recruit, say, Samoans as suicide terrorists in the name of Islam.

Tom West writes:

but who refused to act on it out of political correctness.

Oh come on Steve, surely you can manage better than cheap shots like this.

(1) How many whites does he toss off planes because he doesn't like their look. I expect, to the nearest integer, about zero. In other words, political correctness had *nothing* to do with this.

(2) You know basic stats. Knowing the number of positives is useless without knowing the number of false positives.

The important question is: Freed of political correctness, how many thousands, or tens of thousands, or more would have been barred from flights because they incorrectly triggered someone's "gut instinct".

Without that number, his statement is meaningless as a condemnation of political correctness (if it's even accurate - we both know how often people "knew something was wrong" only after the fact).

I'm obviously not enamored (to put it mildly) of your world-view. But at least you don't normally descend to this sort of Fox news type rhetoric.

Not Blank writes:

Quoth Steve Sailer:

Dr. Jones' argument isn't very good. It is highly costly for al-Qaeda to recruit, say, Samoans as suicide terrorists in the name of Islam.

Fair enough, but they might be able to recruit a Jemaah Islamiyah member or two who won't fit the profile of "nervous Middle Eastern man who smells distinctly of rose water".

Steve Sailer writes:

Nobody remembers Bush's war on ethnic profiling in airport security in his first year in office because Republicans want it shoved down the memory hole and Democrats are against profiling.

But, in the second Presidential debate in 2000, George W. Bush made a big stink over "flying while Arab" -- the Clinton Administration endorsing ethnic profiling in airport security. Bush's goal was to win the Arab vote in Michigan. When Bush got in office, he had his Transportation Secretary Norman Mineta start a campaign to root out ethnic profiling of Muslims.

The guy who checked in Mohammed Atta says he looked fishy told his hometown newspaper:

"Then Tuohey went through an internal debate that still haunts him.

"I said to myself, 'If this guy doesn't look like an Arab terrorist, then nothing does.' Then I gave myself a mental slap, because in this day and age, it's not nice to say things like this," he said. "You've checked in hundreds of Arabs and Hindus and Sikhs, and you've never done that. I felt kind of embarrassed."

It wasn't just Atta's demeanor that caught Tuohey's attention.
"When I looked at their tickets, they had first-class, one-way tickets - $2,500 tickets. Very unusual," he said. "I guess they're not coming back. Maybe this is the end of their trip."

Henry writes:

As Tyler Cowen would say, solve for the equilibrium. The solution is not zero profiling or exclusively monitoring high-risk groups, but monitoring low risk groups just enough so that recruitment costs tend to outweigh the benefits to terrorists of a lower probability of getting caught.

Personally, I'm a fan of Steven Landsburg's idea of profiling but giving generous compensation to the profilees to help alleviate concerns of unfairness.

Ken B writes:

@Justin Talbot: Silas Barta's main point is that recruiting Gramma Frail is hard and very high cost for terrorists. Toss in better profiling for religion and zeal not ethnicity and profiling might be MORE effective at finding white bombers who will mostly be converts. If Gramma Frail recently converted, and was spotted in Death To America parades, she should be flagged. Profiling does not mean *ethnic* profiling. It can be and will be smarter than that.

It's just Bayesian prediction isn't it?

Ken B writes:
Hijacker ethnicity is an endogenous variable
Then why assume profiling will be all about ethnicity?
Marc A Cohen writes:

At ajb:
The Israelis do use profiling successfully - conducted by highly skilled profilers. I don't know how long it would take to adequately train and test the army of profilers it would take to staff all US airports, but I bet it would be a long and expensive process. It might be the case that actual safe flying would double the cost, and that the flying public would in the end prefer to just take their chances.

Mark Bahner writes:

I agree with other commenters that this post is weird. Garrett Jones states that, if profiling was used, Al Qaeda would try to recruit other people outside the profile.

Well, when they recruit senior citizens who have lived in the United States all their lives, let me know. Until then, I'll say that paying equal attention to seniors and males under 40 (particularly those who are not U.S. citizens) is crazy.

I've seen

Keith K. writes:

I agree with Ken B. and Silas Barta on this point.

Also, I see no reason to assume that profiling must be done on an ethnic basis.

Given how data intensive airline flight already is, it would seem to me rather likely that if the airlines were forced to provide their own security that most passengers would be tracked electronically within the flight system.

For example, I would envision the airlines have detailed notes on your travel history. Travel history is (from my understanding) a huge red-flag concerning attack-likelihood in most cases for a given passenger. If your tracked making a bunch of flights between europe and pakistan for example and there is no business reason for this to occur, this strikes me as exactly the sort of thing that would red-flag you in the system when you try to buy you airline ticket. You would then be forced to undergo additional processing when you get to the airport.

There are probably numerous other actuarial techniques which could be employed along this line. Let the market work, and the market will find them.

Ken B writes:

Not Blank:

but they might be able to recruit a Jemaah Islamiyah member or two who won't fit the profile

Any other profile they might fit?

Chunlin writes:

"And if ethnic profiling became legal, terrorist organizations would have an even stronger incentive to recruit from non-profiled demographic groups." Yes,that's true. In fact, the profiling will continue to work no matter if the government will credibly commits to treating everyone identically or switches to 100% profiling, because that's important.

Mark Bahner writes:

My lunchtime comments somehow got cut off, but what I was going to say was that I've seen a child clutching a security blanket (or maybe it was stuffed animal) traveling with her mother and grandmother, and they end up getting three times as much attention as the strapping man who went through right after them. (Because the child, mother, and grandmother each got equal time spent on them as the strapping man.) It simply doesn't make sense.

Let's look at all the attacks or attempted attacks on airliners in the U.S. in the whole history of flight in the U.S. I doubt a single one has been by a woman, child, or man over 60. Or come to think of it, a man traveling with his family.

So it's simply good economics to spend less time on women, children, men over 60, and men traveling with their families.

Tom West writes:

Let's look at all the attacks or attempted attacks on airliners in the U.S. in the whole history of flight in the U.S. I doubt a single one has been by a woman, child, or man over 60. Or come to think of it, a man traveling with his family.

This sounds like every general who lost a war preparing to fight the last battle. "We know how they work, and this will stop them." Unfortunately, it's actually we *knew* how they worked and we've made ourselves defenseless when they changed.

However, if you are interested in security theater (i.e. useless measures that make passengers *feel* safer) then profiling is definitely the way to go. Certainly I think private security would tend to go that way unless actually curbed.

However, in all of this, may I point out that we've neglected to price in how many new home grown terrorists we produce by profiling. Given how enraged I've seen people become when they've been "treated like criminals" once, I can only imagine the effect when it happens consistently, to say nothing of the effect that seeing a visible group treated like criminals has upon reinforcing incorrect perceptions.

Ken B writes:

Tom West:

reinforcing incorrect perceptions

Which incorrect perceptions are those? That frail old ladies are less able to overpower pilots than healthy young men? That middle aged Quakers travelling with their children are less likely to be suicide bombers than recent Wahabi converts?

Profiling is about a lot more than just ethnicity. Good profiling will recognize that many of middle eastern descent are in fact not threats at all.

Ken B writes:

Most of Garett's objections to profiling -- and several others' objections too -- carry over to facial recognition. After all we can use FR to recognize known individuals who are known to have worrying pasts or connections. Would we eschew using FR because, hey they'll recruit new people? Of course not.

I was profiled this week in fact. I bought new auto insurance. They cahecked my driving history, my car's history, my credit score, my address, my age, aspects of my health history. Who knows what else; I know they had my phone number from 1998. And they did it in a blink without me feeling a thing.

Mark Bahner writes:
This sounds like every general who lost a war preparing to fight the last battle.

Would you like to bet on this assessment? If the next attack or attempted attack on a U.S. airliner is by a person or group of people over 60 (including, of course, Hell's Grannies) then I'll pay you $100. On the other hand, if the next attack or attempted attack (e.g., publicly reported plot) on a U.S. airliner is by men under 60, you pay me $20. And if it involves Muslim men under 60, you pay me $40. How about it?

However, if you are interested in security theater (i.e. useless measures that make passengers *feel* safer) then profiling is definitely the way to go.

B.S. It's simple math and economics. As Ken B. points out directly above, profiling is the way to go if you're interested in actually providing security (or insurance), rather than security theater (or insurance theater). Security theater is why the TSA spends the same amount of effort on each senior citizen (man and woman!) as they do on every Muslim man under 40.

It's simple math and economics that, if a particular person is 1000 times more likely to commit a particular crime, one should spend 1000 times more effort on checking that person out than each person that's 1000 times less likely to commit a particular crime. That's why, when the police find a forged check, they don't go to the local kindergarten to check out possible suspects.

The TSA is involved in security theater. That's why everyone must get checked equally, because many members of the audience would "boo" if everyone wasn't checked equally.

However, in all of this, may I point out that we've neglected to price in how many new home grown terrorists we produce by profiling.

The question is whether the additional expense involved with "new home grown terrorists" produced by profiling is greater than the present expense of searching senior citizens and children equally to Muslim men under 40. Especially Muslim men under 40 who have lived outside the U.S. Especially Muslim men under 40 who have lived outside the U.S. and aren't flying on business. Especially Muslim men under 40 who have lived outside the U.S. and aren't flying on business.

Considering that the TSA is expected to cost about $8 billion in 2012, and probably 99.9+ percent of that spending is NOT on checking out Muslim men under 40 who have lived outside the U.S. and who aren't flying on business, I'm guessing that more money would be saved by profiling.

A great way to see what works and what doesn't would be to get rid of the TSA and return to private security at airports. It's amazing how a profit motive returns theater back to where it belongs...in theaters.


John T. Kennedy writes:

Garett,

"However, if the government (or a private security firm) switched to 100% profiling, then it's obvious what the terrorist organization will do: Find someone outside the profile."

Why not sidestep the the profiling completely and attack a commuter bus, movie theater, or busy restaurant instead of an airport crawling with security?

Seems obvious to me that if you harden airport security the terrorist organization would simply attack softer targets - if the organization truly existed as a serious threat.

Thus I conclude it unlikely that such serious threats exist.

John David Galt writes:

A truly private (unregulated) airline industry would give the tiny danger of hijacking the level of precautions it deserves: zero. Falling coconuts kill more people each year than all terrorism outside of the Middle East, much less airline terrorism (which stopped working *during* 9/11, in time to prevent Flight 93 reaching the bad guys' target).

Meanwhile, the TSA kills more people each month than died on 9/11, simply by annoying you and me into driving rather than flying. And in the process, they're deliberately conditioning the public to put up with abuses by police (and imitation police such as themselves).

The TSA must be abolished, and until it is, the terrorists have won. No lesser reform will help.

Comments for this entry have been closed
Return to top