| Homeowners engage in private precaution when public protection is inadequate. When this precaution is observable to criminals, as in the case of a home burglar alarm system advertised with a sign or sticker, it has the beneficial effect of decreasing the probability of burglary of alarmed homeowners. However, because a burglar is rational in his criminal activity, he will avoid protected homes and target homeowners without alarms, increasing their probability of burglary. Observable private precaution therefore incorporates both the deterrence (Clotfelter 1978) and diversion (Shavell 1991) effects. Even though burglar alarms have been identified as effective deterrents, little research has been devoted to determining if they also divert crime. In this dissertation, I examine the diversion effect associated with burglar alarms. In chapter three, I estimate the net effect, or combined deterrence and diversion effects, of burglar alarms on burglary rates and find that burglary rates fall only slightly with increases in burglar alarm use. Assuming that burglar alarms deter, this finding suggests the presence of the diversion effect muting deterrence. In chapters four and five, I present two methods for measuring the diversion effect associated with burglar alarms using data from the homeowner's insurance industry. Since companies with more non-alarmed customers face larger costs due to diversion, chapter four measures the diversion effect as the relationship between the market shares of homeowner's insurance companies and the protective device discounts they offer. Chapter five's approach is based on the empirical method of Berry, Levinsohn, and Pakes (1995) and uses supply and demand parameters in an oligopolistic framework to measure the diversion effect as the difference in the probabilities of burglary of non-alarmed homeowners when some homes install alarms and when no homes are alarmed. Both methods show that diversion can be measured. When empirically tested, they also yield estimates of the diversion effect. Although the social effects associated with burglar alarms are likely to be positive on balance, I expose the hidden costs that may be associated with observable precautionary measures. |
