When your neighbor installs a burglar alarm, thoughtful burglars are encouraged to choose a different target--like your house, for example. It's rather as if your neighbor had hired an exterminator to drive all the vermin next door. On the other hand, if your neighbor installs video cameras that monitor the street in front of both your houses, he might be doing you a favor. So the spillover effects of self-protection can be either good or bad.
Consider the different ways that people self-protect against car theft. Devices like alarm systems and the "Club" have a social upside: Their proliferation might make car theft so unprofitable that potential thieves would decide to seek more useful employment (though, on the other hand, it's possible that they'll seek employment as, say, arsonists or killers for hire). But those same devices have a social downside: They encourage thieves to prey more heavily on those who haven't bought one. From a social viewpoint, if the total number of thefts does not change, then the expenditure on alarm systems is pure waste.
For a much lower cost, you can install "fake" self-protection--say, a little blinking red light that looks like it's attached to an alarm system, or a cheap piece of foam rubber that looks from a distance like the heavy metal Club. Here again you're imposing a cost on your neighbors: If these devices become common, the value of the real thing is diluted.
That point was driven home to me the last time I shopped for a car. Acura offered a security system as mandatory equipment. Toyota allowed you to buy a car without a security system. You could then go out and install your own system for considerably less than what Acura was (implicitly) charging.
But I decided that Acura's system--even at a much higher price--was the better deal. Professional car thieves know that the security system is mandatory on an Acura, and therefore know that my blinking red light is for real. With the Toyota, even if I do install a real security system, thieves might suspect me of trying to fool them and smash my windows to find out.
There's another kind of security system, available only in a few cities. The "Lojack" is a hidden radio transmitter that can be activated after your car is stolen, to lead police to the thief (or, better yet, to the chop shop that employs the thief). The transmitter is hidden randomly within the car, so thieves cannot easily find it and deactivate it.
The Lojack is completely hidden. There's no way to look at a car and know whether it has a Lojack installed. So unlike, say, the Club, a Lojack will never prevent any particular car from being stolen; it will only increase the chance of its being recovered. But from a social point of view, the Lojack has the huge advantage of helping your neighbors rather than hurting them. The Club convinces thieves to steal someone else's car instead; the Lojack convinces thieves not to steal.
And it does so with remarkable effectiveness. Economists Ian Ayres and Steven Levitt have examined the effects of the Lojack in about a dozen cities over the past 10 years (its first introduction was in Boston in 1986). Their task wasn't easy, because just as the prevalence of the Lojack affects auto-theft rates, so auto-theft rates affect the prevalence of the Lojack--first because consumers buy more security equipment when theft rates are high, and second because regulators behave differently when thefts are high.
But after sorting all this out, Ayres and Levitt found that the Lojack has an astoundingly large effect on auto-theft rates. It turns out that a 1 percent increase in Lojack sales can reduce auto-theft rates by 20 percent or more. What's happening to all those car thieves? Are they moving to other cities, or are they becoming house burglars, or are they turning into socially useful citizens? Ayres and Levitt examined these difficult questions also, and their bottom-line conclusion is that the Lojack really does prevent a lot of crime, rather than just moving it to other venues.
In fact, although it costs only about $100 a year to have a Lojack, Ayres and Levitt estimate that each individual Lojack prevents about $1,500 a year in losses due to theft. In most cases, that $1,500 benefit accrues not to the Lojack owner, but to strangers.
By the criteria that economists usually employ, this suggests that Lojacks should be heavily subsidized, just as visible security systems--like my neighbor's home burglar alarm or the Club--should be taxed. When you're doing something that makes strangers better off, you should be encouraged to do more of it.
If we all used the same insurance company, you might expect that company to supply the appropriate subsidy. As long as your Lojack reduces the number of insurance claims, the company should be willing to pay you to install it. But with multiple insurance companies, that doesn't work so well: A company that insures only 10 percent of the populace will reap only 10 percent of the Lojack's benefits, and so will undersubsidize them. Worse yet, large insurance discounts are illegal in many states.
The media have recently paid a lot of attention to research on other kinds of self-protection, most notably the work of John Lott and David Mustard on concealed handguns. But the Lojack research is in many ways more informative, because the authors were able to do a thorough job of distinguishing between benefits to the purchaser of a Lojack and benefits to the community at large. That discrepancy is the sort of thing that leads markets to fail--in this case by providing too many Clubs and not enough Lojacks.