The role of handguns in crime has been the subject of extensive policy and academic debate in recent years. The interest in the issue has grown with the enactment of the restrictive (federal) Brady Bill and the adoption by many states of the right-to-carry concealed-handgun laws. These “shall issue” laws make it much easier for noncriminals to obtain licenses to carry concealed handguns. (Ten states passed such laws from 1977 to 1992, and 13 states have passed such laws since 1992.) This observed dichotomy in policy reflects the lack of consensus among policymakers regarding the role of handguns in violence. A similar disagreement exists in academic circles. For example, results reported in Philip Cook et al. (1995), Arthur Kellermann et al. (1995), Cook and Jen Ludwig (1996), and David Hemenway (1996) imply that an increase in gun ownership increases rates of crime. Daniel Polsby (1995) and John Lott and David Mustard (1997) are doubtful of such implications.

There are two theoretical possibilities regarding the effects of these laws. Increased gun ownership might lead to increased crime because of the increased availability of guns – the facilitating effect. Alternatively, because increased gun ownership might help potential victims to arm and protect themselves, thus increasing the criminals’ uncertainty regarding an armed response, such laws might lead to reduced crime against persons-the deterrent effect. Which effect dominates might depend on population characteristics in a given jurisdiction. Concealed-handgun laws might lead to increases in crime in some jurisdictions, while leading to decreases in other jurisdictions. In a controversial paper, Lott and Mustard (1997) have examined this issue. They find that passage of concealed-handgun (shallissue) laws by a state causes a significant reduction in violent as well as property crimes (Lott and Mustard, 1997 table 11). They attribute the results to a deterrent effect: as criminals become aware that victims might be armed, they reduce the rate of commission of crimes.

We believe Lott and Mustard’s findings are suspect, mainly because of the way they parameterize and measure the effect of permissive handgun laws on crime. They model the effect as a shift in the intercept of the linear crime equation they estimate at the county level. This approach is predicated on two assumptions: (i) all behavioral (response) parameters of this equation (slope coefficients) are fixed (unaffected by the law), and (ii) the effect of the law on crime is identical across counties. Obviously, if the law affects the behavior of the criminals or of citizens, then these response parameters should change, and not only the intercept term. Moreover, it seems highly unlikely that the magnitude of the effects such laws may have on crime rates in a county would be independent of economic and demographic characteristics of the county. In fact, the effect may vary with the age and gender composition of the population and the economic conditions of the counties, among other things. Others who have commented on Lott and Mustard (1997) (e.g., Ludwig, 1996; Dan Black and Daniel Nagin, 1998) have overlooked this problem, and consequently their alternative estimates may be subject to the same criticism; see also the reply by Lott (1998) to Black and Nagin’s criticisms.

We reexamine the effect of permissive concealed-handgun laws on crime using a procedure to overcome these shortcomings, allowing all behavioral parameters of the model to change. Our method also allows the effect of the law on crime rates to be heterogeneous across counties so that we can infer how various factors influence the magnitude of the change in crime resulting from such laws. More specifically, we project the 1992 crime rate for counties without such laws if they had adopted the law by 1992. We then compare these projections, which are a function of county characteristics, with actual crime data for 1992 to infer how the absence of the law has affected crime in these counties. We also examine the relationship between these projected changes and county characteristics.