More Guns Less Crime
Page 10
CONCEALED-HANDGUN LAWS AND CRIME RATES/83
(1) adding a variable to control for state laws that increase sentencing penalties when crimes involve guns and (2) adding variables to measure the impact of waiting periods. 38 It is not clear whether adding an extra day to a waiting period had much of an effect; therefore, I included a variable for when the waiting period went into effect along with variables for the length of the waiting period in days and the length in days squared to pick up any differential impact from longer lengths. In both sets of regressions, the variable for nondiscretionary concealed-handgun laws remains generally consistent with the earlier results. 39 While the coefficients for arrest rates are not reported here, they also remain very similar to those shown previously.
So what about these other gun laws? The pattern that emerges from table 4.11 is much more ambiguous. The results for county-level data suggest that harsher sentences for the use of deadly weapons reduce violent crimes, especially crimes of aggravated assault and robbery. While the same county-level data frequently imply an impact on murder, rape, aggravated assault, and robbery, the effects are quite inconsistent. For example, simply requiring the waiting period appears to raise murder and rape rates but lower the rates for aggravated assault and robbery. The lengths of waiting periods also result in inconsistent patterns: longer periods at first lower and then raise the murder and rape rates, with the reverse occurring for aggravated assault. Using state- level data fails to confirm any statistically significant effects for the violent-crime categories. First, it reveals no statistically significant or economically consistent relationship between either the presence of waiting periods or their length and violent-crime rates. The directions of the effects also differ from those found using county data. Taken together, the results make it very difficult to argue that waiting periods (particularly long ones) have an overall beneficial effect on crime rates. In addition, one other finding is clear: laws involving sentence length and waiting periods do not alter my earlier findings with respect to nondiscretionary laws; that is, the earlier results for nondiscretionary laws cannot merely be reflecting the impact of other gun laws.
The Importance of the Types of Concealed-Handgun Laws Adopted: Training and Age Requirements
Finally, we need to consider how concealed-handgun laws vary across states and whether the exact rules matter much. Several obvious differences exist: whether a training period is required, and if so, how long that period is; whether any minimum age limits are imposed; the number of
Table 4.11 Controlling for other gun laws
State-level regressions
Nondiscretionary law
adopted Enhanced sentencing
law adopted Waiting law adopted Percent change in crime
by increasing the
waiting period by one
day: linear effect Percent change in crime 0.12% -0.13% 0.59%* -0.041% 0.59%** -0.021% 0.05% -0.06% -0.25%
by increasing the
wating period by one
day: squared effect
Note: The control variables are the same as those used in table 4.1, including year and county dummies, though they are not reported, because the coefficient estimates are very similar to those reported earlier. All regressions use weighted least squares, where the weighting is each county's population. *The result is statistically significant at the 1 percent level for a two-tailed t-test. **The result is statistically significant at the 5 percent level for a two-tailed t-test. ***The result is statistically significant at the 10 percent level for a two-tailed t-test.
86/CHAPTER FOUR
years for which the permit is valid; where people are allowed to carry the gun (for example, whether schools, bars, and government buildings are excluded); residency requirements; and how much the permit costs. Six of these characteristics are reported in table 4.12 for the thirty-one states with nondiscretionary laws.
A major issue in legislative debates on concealed-handgun laws is whether citizens will receive sufficient training to cope with situations that can require difficult, split-second decisions. Steve Grabowski, president of the Nebraska state chapter of the Fraternal Order of Police, notes that "police training is much more extensive than that required for concealed-handgun permits. The few hours of firearms instruction won't prepare a citizen to use the gun efficiently in a stress situation, which is a challenge even for professionals." 40 Others respond that significantly more training is required to use a gun offensively, as a police officer may be called on to do, than defensively. Law-abiding citizens appear reticent to use their guns and, as noted earlier, in the majority of cases simply brandishing the gun is sufficient to deter an attack.
Reestimating the earlier regressions, I included measures for whether a training period was required, for the length of the training period, and for the age limit. 41 The presence or length of the training periods typically show no effect on crime, and although the effects are significant for robbery, the size of the effect is very small. On the other hand, age limits display quite different and statistically significant coefficients for different crimes. The 21-year-old age limit appears to lower murder rates, but it tends to reduce the decline in rape and overall violent-crime rates that is normally associated with nondiscretionary concealed-handgun laws. Because of these different effects, it is difficult to draw firm conclusions regarding the effect of age limits.
Recent Data on Crime Rates
After I originally put the data together for this study, and indeed after I had written virtually all of this book, additional county-level data became available for 1993 and 1994 from the FBI's Uniform Crime Reports. These data allow us to evaluate the impact of the Brady law, which went into effect in 1994. Four additonal states (Alaska, Arizona, Tennessee, and Wyoming) also had right-to-carry laws in effect for at least part of the year. The new information allows us to double-check whether the results shown earlier were mere aberrations.
Table 4.13 reexamines the results from tables 4.1, 4.8, and 4.11 with these new data, and the findings are generally very similar to those already reported. The results in section A that correspond to table 4.1 imply an even larger drop in murder rates related to the passage of concealed-handgun laws (10 percent versus 7.7 percent previously), though
Table 4.12 Current characteristics of different nondiscretionary concealed-handgun laws
State with
nondiscretionary
law
Last
significant
modification
Permit
duration
(years)
Training
length
(hours)
Age
requirement
Initial fee
Renewal fee
Issuing agency
Alabama Alaska
Arizona
Arkansas Connecticut Florida Georgia
Idaho Indiana
Kentucky
Louisiana
1936 1994
1995
1995 1986 1995 1996
1996 1980
1996
1996
$15-25 $123
$50
$100 $35 $85 $32
$56
$25
$60
$100
$6
$57
$35 $70
$24 $15
$60
Sheriff
Dept. of Public
Safety
Dept. of Public
Safety
State Police
State Police
Dept. of State
Judge of
Probate Court
Sheriff
Chief of Police
or Sheriff
Sheriff
Dept. of Public Safety
Table 4.12 Continued
■This training period is waived for those who receive a permit directly from their local sheriff. b The fee is reduced to $70 for those who are over 60
years of age.
Table 4.13 Earlier results reexamined using additional data for 1993 and 1994
Percent change in various crime rates for changes in explanatory variables
Change in explanatory Violent Aggravated
variable crime Murder Rape assault Robbery
Section A: -4.4%* -10.0%* -3.0%* -5.7%* 0.6%
Nondiscretionary law adopted
Section B: The difference -0.5%* -2.9%* -1.7%* -0.3%* -2.2%*
in the annual change in crime rates in the years before and after the change in the law (annual rate after the law minus annual rate before the law)
Section C: Brady law 3% -2.3% 3.9%** 3.7%** -3.9%
adopted
Note: This table uses county-level, violent-crime data from the Uniform Crime Report that were not available until the rest of the book was written. Here I was not able to control for all the variables used in table 4.1. All regressions use weighted least squares, where the weighting is each county's population. Section C also controls for the other variables that were included in Table 4.11 to account for changes in other gun laws. Section A corresponds to the regressions in table 4.1, section B to those in table 4.8, and section C to those in table 4.11, except that a dummy variable for the Brady law was added for those states that did not previously have at least a five-day waiting period.
*The result is statistically significant at the 1 percent level for a two-tailed t-test. **The result is statistically significant at the 10 percent level for a two-tailed t-test.
the declines in the rates for overall violent crime as well as rape and aggravated assault are smaller. Robbery is also no longer statistically significant, and the point estimate is even positive. As noted earlier, given the inverted V shape of crime-rate trends over time, comparing the average crime rates before and after the passage of these laws is not enough, since crime rates that are rising before the law and falling afterward can produce similar average crime rates in the two periods. To deal with this, section B of table 4.13 corresponds to the results reported earlier in table 4.8. The estimates are again quite similar to those reported earlier. The effect on rape is larger than those previously reported, while the effects for aggravated assault and robbery are somewhat smaller. All the results indicate that concealed-handgun laws reduce crime, and all the findings are statistically significant.
Finally, section C of table 4.13 provides some very interesting estimates
CONCEALED-HANDGUN LAWS AND CRIME RATES/91
of the Brady law's impact by using a variable that equals 1 only for those states that did not previously have at least a five-day waiting period. The claims about the criminals who have been denied access to guns as a result of this law are not necessarily evidence that the Brady law lowers crime rates. Unfortunately, these claims tell us nothing about whether criminals are ultimately able to obtain guns illegally. In addition, to the extent that law-abiding citizens find it more difficult to obtain guns, they may be less able to defend themselves. For example, a woman who is being stalked may no longer be able to obtain a gun quickly to scare off an attacker. Numerous newspaper accounts tell of women who were attempting to buy guns because of threats by former lovers and were murdered or raped during the required waiting period. 42
The evidence from 1994 indicates that the Brady law has been associated with significant increases in rapes and aggravated assaults, and the declines in murder and robbery have been statistically insignificant. All the other gun-control laws examined in table 4.11 were also controlled for here, but because their estimated impacts were essentially unchanged, they are not reported.
What Happens to Neighboring Counties in Adjacent States When Nondiscretionary Handgun Laws are Adopted?
If you put more resources in one place, it will displace some of the crime.
Al LVEcuyer, West Boylston (Massachusetts) Police Chief 3
Up to this point we have asked what happens to crime rates in places that have adopted nondiscretionary laws. If these laws do discourage criminals, however, they may react in several ways. We already have discussed two: criminals could stop committing crimes, or they could commit other, less dangerous crimes like those involving property, where the probability of contact with armed victims is low. Yet as the epigraph for this section notes, a third possibility is that criminals may commit crimes in other areas where potential victims are not armed. A fourth outcome is also possible: eliminating crime in one area can help eliminate crime in other areas as well. This last outcome may occur if criminals had been using the county that adopted the law as a staging area. Crime-prone, poverty-stricken areas of cities may find that some of their crime spills over to adjacent areas.
This section seeks to test what effect concealed-handgun laws and
higher arrest rates have on crime rates in adjacent counties in neighboring states. Since concealed-handgun laws are almost always passed at the state level, comparing adjacent counties in neighboring states allows us to examine the differential effect of concealed-handgun laws. Evidence that changes in a state's laws coincide with changes in crime rates in neighboring states will support the claim that the laws affect criminals. If these laws do not affect criminals, neighboring states should experience no changes in their crime rates.
Although any findings that nondiscretionary concealed-handgun laws cause criminals to leave the jurisdictions that adopt these laws would provide additional evidence of deterrence, such findings would also imply that simply looking at the direct effect of concealed-handgun laws on crime overestimates the total gain to society from these laws. In the extreme, if the entire reduction in crime from concealed-handgun laws was simply transferred to other areas, society as a whole would be no better off with these laws, even if individual jurisdictions benefited. While the evidence would confirm the importance of deterrence, adopting such a law in a single state might have a greater deterrent impact than if the entire nation adopted the law. The deterrent effect of adopting nondiscretionary concealed-handgun laws in additional states could also decline as more states adopted the laws.
To investigate these issues, I reran the regressions reported in table 4.1, using only those counties that were within fifty miles of counties in neighboring states. In addition to the variable that examines whether your own state has a nondiscretionary concealed-handgun law, I added three new variables. One variable averages the dummy variables for whether adjacent counties in neighboring counties have such laws. A second variable examines what happens when your county and your neighboring county adopt these laws. Finally, the neighboring counties' arrest rates are added, though I do not bother reporting them, because the evidence indicates that only the arrest rates in your own county, not your neighboring counties, matter in determining your crime rate.
The results reported in table 4.14 confirm that deterrent effects do spill over into neighboring areas. For all the violent-crime categories, adopting a concealed-handgun law reduces the number of violent crimes in your county, but these results also show that criminals who commit murder, rape, and robbery apparently move to adjacent states without the laws. The one violent-crime category that does not fit this pattern is aggravated assault: adopting a nondiscretionary concealed-handgun law lowers the number of aggravated assaults in neighboring counties. With respect to the benefits of all counties adopting the laws, the last column
CONCEALED-HANDGUN LAWS AND CRIME RATES/93
Table 4.14 Estimates of the impact of nondiscretionary concealed-handgun laws on neighboring counties
shows that all categories of violent crime are reduced the most when all counties adopt such laws. The results imply that murder rates decline by over 8 percent and aggravated assaults by around 21 percent when a county and its neighbors adopt concealed-handgun laws.
As a final test, I generated the figures showing crime trends before and after a neighbor's adoption of the law by the method previously used, in addition to the time trends for before and after one's own adoption of the concealed-handgun law
s. The use of an additional squared term allows us to see if the effect on crime is not linear. Figures 4.10—4.13 provide a graphic display of the findings for the different violent-crime categories, though the results for the individual violent-crime categories are equally dramatic. In all violent-crime categories, the adoption of concealed-handgun laws produces an immediate and large increase in violent-crime rates in neighboring counties, and in all the categories except aggravated assaults the spillover increases over time just as the counties with the nondiscretionary law see their own crime rates continue to fall. The symmetry and timing between the reduction in counties with nondiscretionary laws and increases in neighboring counties without the laws is striking.
Overall, these results provide strong additional evidence for the deterrent effect of nondiscretionary concealed-handgun laws. They imply that the earlier estimate of the total social benefit from these laws may have overestimated the initial benefits, but underestimated the long-term
-2-101234 Years before and after the neighbor's adoption of the law
Figure 4.10. Impact on murder rate from a neighbor's adoption of nondiscretionary concealed-handgun law
-2-10 1 2 3 4
Years before and after the neighbor's adoption of the law
Figure 4.11. Impact on robbery rate from a neighbor's adoption of nondiscretionary concealed-handgun law
t 1 3g ' H 1 1 1 r
-2-101234 Years before and after the neighbor's adoption of the law
Figure 4.12. Impact on rape rate from a neighbor's adoption of nondiscretionary concealed-handgun law
i l i i
-10 12 3 4
Years before and after the neighbor's adoption of the law
Figure 4.13. Impact on aggravated assault rate from a neighbor's adoption of nondiscretionary concealed-handgun law
benefits as more states adopt these laws. In the long run, the negative spillover effect subsides, and the adoption of these laws in all neighboring states has the greatest deterrent effect on crime.