Research Summary
Brookings Institution Center on Urban & Metropolitan Policy
1775 Massachusetts Ave NW
Washington DC 20036
Phone: 202-797-6139
http://www.brookings.edu/urbanGuns have both virtuous and vicious uses. Gun use by criminal assailants increases the scope and lethality of violent crime. On the other hand, gun use by potential victims can scare off assailants and prevent injury. The balance between virtuous and vicious uses has traditionally favored keeping a gun at home over carrying one in public, with the latter subject to more stringent regulation. Guns in the home do not directly threaten the public at large and may enhance the capacity for defending against intruders. Furthermore, armed households arguably provide a deterrent to residential burglary, and particularly to burglaries of occupied dwellings, or "hot" burglaries.
The deterrence may also extend beyond the household. If burglars lack "inside" knowledge about which households are armed, this crime-control benefit can extend to the entire community. Recently, we conducted the first systematic study of the claim that guns deter hot burglaries. Our conclusion: Increasing the prevalence of guns in a community may, if anything, slightly increase the chance of burglary victimization, and has no effect on hot burglaries. One possible explanation for the findings is that guns are valuable "loot," and that gun-rich communities are especially profitable to burglars.
Guns and Burglary: Previous "Evidence"
Advocates assert the idea that guns deter burglary frequently and with great confidence. However, the empirical support for this assertion is surprisingly weak. The available evidence consists of anecdotes, interviews with burglars, casual international comparisons, and some hard-to-interpret results from surveys. National victimization surveys have found that the frequency of gun use in self-defense ranges from as high as 503,000 incidents in the preceding year to as low as 32,000 incidents, depending on how the survey is conducted and what questions are asked.
In one sample of state prisoners, 74 percent agreed that one reason burglars avoid residences when people are at home is that they fear being shot. At the same time, burglars also report that guns are of considerable value to them. Nearly half of the respondents in the prison survey indicated that they had stolen a gun during their lifetimes. Interviews with active burglars suggest that they typically prefer items that are easy to carry, easily concealed, and have high "pound for pound" value. As one burglar reported: "A gun is money with a trigger".
The most influential evidence in support of a burglary-deterrent effect comes from crossnational differences in burglary patterns. In the United Kingdom, where relatively few households have guns, crime victimization surveys suggest that 45 percent or more of burglaries are of occupied dwellings. Compared with the U.K., the proportion of "hot" burglaries in the United States is much lower and guns in the house much more common. However, American and British households differ in a number of other ways beyond gun ownership that are likely to affect the cost-benefit calculus facing burglars; home invasion burglars in Britain face a much more lenient prison sentence if caught, and households in Britain are less likely to have a dog or a man living in them. Without controlling for these other differences that may be important, it is difficult to attribute the disparity in hot burglary rates to gun prevalence.
Better Evidence
Our review provides the first systematic evidence on the effects of guns on hot burglary by comparing burglary patterns across states and counties within the United States. Unlike previous international comparisons, our analysis explicitly attempts to address many of the other ways in which counties and states within the United States may differ from one another beyond gun prevalence. Any such unmeasured differences across areas can bias the results, leading analysts to either understate or overstate the effects of gun ownership on crime. The analysis is made possible by a new geo-coded version of the National Crime Victimization Survey (NCVS) for the years 1994 through 1998, which provides the only nationally representative information about hot burglaries in the United States. The NCVS collects victimization reports from 50,000 to 60,000 households each year, and surveys every household resident age 12 and older for a total of 90,000 to 100,000 survey respondents. The response rate for the NCVS is typically about 95 percent. The authors merge information from the NCVS with a measure of gun prevalence in the respondent's county.
Complementary analyses are conducted using annual county- or state-level data on crimes reported to the police, as recorded by the FBI's Uniform Crime Reports (1977-1998). Unfortunately, the UCR data, unlike the NCVS, do not distinguish between burglaries of occupied and unoccupied buildings. Because the United States does not maintain a registry of guns in private hands, and surveys do not provide data for each of the 50 states, the authors use a proxy variable--the percentage of suicides committed with a gun--to explore the effect of gun prevalence at the state and county levels. The percent of suicides with guns has been shown to be a reliable proxy, outperforming such measures as the percentage of homicides committed with a gun, the prevalence of membership in the National Rifle Association, or subscription rates to gun-oriented magazines.
Findings
Results from the Uniform Crime Reports and NCVS data suggest that gun prevalence may increase the odds of being burglarized. Our analyses find that a 10-percent increase in a county's gun ownership rate is associated with a 3-percent to 7-percent increase in the likelihood that a home will be burglarized. The likelihood that a burglary is "hot" (that is, that a resident is at home during the break-in) does not appear to be affected by the prevalence of gun ownership. The finding that gun prevalence increases the odds of burglary appears to be fairly robust no matter what the choice of data or method.
The main findings from the NCVS come from a cross-section regression that controls for a detailed set of individual- and household-level characteristics and, in some specifications, local-area crime rates and socio-demographic measures. Omitted state-level policy actions or social factors do not seem to explain the findings since similar results hold when the authors focus only on within-state, across-county comparisons. Finally, analysis of panel data from the UCR yields similar findings, even after controlling for unmeasured county-level variables. Testing the possibility of reverse causation with the UCR panel, we find that gun prevalence drives burglary, but burglary does not drive gun prevalence.
As a final check on the possibility that these results are driven by unmeasured, confounding factors, we use an instrumental-variables approach to isolate variations of gun prevalence related to differences in each area's rural tradition (as measured by the proportion of the state that was rural in 1950). Because rural areas, on average, have less crime and more guns than other jurisdictions, this procedure is likely to overstate the "deterrence" effects of guns on burglary and understate any effect that guns may have on increasing burglary rates. The analysis still yields a positive relationship between guns and burglary, and as a result, one may be fairly confident that the true relationship is not negative. One possible reason why the risk of burglary increases with gun prevalence is that guns are valuable loot. Providing some support for this theory is the fact that in 14 percent of the burglaries in the NCVS data in which a gun was stolen, it was the only item stolen. On the other hand, the likelihood of a gun being stolen in a residential burglary in the NCVS is just 5 percent. An increase of 10 percentage points in gun prevalence would therefore increase by 1 percentage point the probability that a gun would be stolen during a burglary, which would increase the average payoff to a burglary by 5 percent and the median burglary value by 20 percent.
Conclusion
Keeping a gun at home is unlikely to provide a net benefit to the rest of the community in the form of burglary deterrence. If anything, residences in a neighborhood with more guns may be at greater risk of being burglarized. The upshot is ironic: Guns are often kept to protect the home, but the aggregate effect of keeping guns at home may be to increase the victimization rate.
Source: "Guns and burglary" by Philip Cook and Jens Ludwig, in: Evaluating Gun Policy, Brookings Institution Press, 2003.
