More Guns Less Crime
Page 9
70/CHAPTER FOUR
specials." Indeed, the results have implications for many gun-control rules that raise gun prices. Law-abiding minorities in the most crime-prone areas produced the greatest crime reductions from being able to defend themselves. Unfortunately, however unintentionally, California's new laws risk disarming precisely these poor minorities.
Using Other Crime Rates to Explain the Changes in the Crime Rates Being Studied
Other questions still exist regarding the specifications employed here. Admittedly, although arrest rates and average differences in individual counties are controlled for, more can be done to account for the changing environments that determine the level of crime. One method is to use changes in other crime rates to help us understand why the crime rates that we are studying are changing over time. Table 4.5 reruns the specifications used to generate figure 4.1 A but includes either the burglary or robbery rates as proxies for other changes in the criminal justice system. Robbery and burglary are the violent- and property-crime categories that are the least related to changes in concealed-handgun laws, but they still tend to move up and down together with all the other types of crimes. 33
Some evidence that burglary or robbery rates will measure other changes in the criminal justice system or other omitted factors that explain changing crime rates can be seen in their correlations with other crime categories. Indeed, the robbery and burglary rates are very highly correlated with the other crime rates. 34 The two sets of specifications reported in table 4.5 closely bound the earlier estimates, and the estimates continue to imply that the introduction of concealed-handgun laws coincided with similarly large drops in violent crimes and increases in property crimes. These results differ from the preceding results in that the nondiscretionary laws are not significant related to robberies. The estimates on the other control variables also remain essentially unchanged. 35
Crime: Changes in Levels Versus Changes in Trends
The preceding results in this chapter examined whether the average crime rate fell after the nondiscretionary laws went into effect. If changes in the law affect behavior with a lag, changes in the trend are probably more relevant; therefore, a more important question is, How has the crime trend changed with the change in laws? Examining whether there is a change in levels or a change in whether the crime rate is rising or falling could yield very different results. For example, if the crime rate
Table 4.5 Using crime rates that are relatively unrelated to changes in nondiscretionary laws as a method of controlling for other changes in the legal environment: controlling for robbery and burglary rates
Table 4.5 Continued
Percent change in various crime rates for changes in explanatory variables
Violent Aggravated
crime Murder Rape assault
Change in the explanatory variable
Robbery
Property crime
Burglary Larceny
Auto theft
Controlling for burglary rates
Nondiscretionary law adopted multiplied by *county population (evaluated at mean county population)
Arrest rate for the crime category increased by 100 percentage points
-2.4%* 1%
-0.026* 5%
-4.3%* 1.1%
-0.13* 6%
-2.0%* 0.4%
-0.05*
3%
-2.6%* 0.4%
-0.05* 5%
0.4% 0.04%
-0.043* 3%
1.8%* 0.7%
-0.05* 6%
1.4%* 0.4%
-0.01* 2%
3.6%* 0.5%
-0.01* 2%
Note: While not all the coefficient estimates are reported, all the control variables are the same as those used in table 4.1, including year and county dummies. All regressions use weighted least squares, where the weighting is each county's population. Net violent and property-crime rates are respectively net of robbery and burglary rates to avoid producing any artificial collinearity. Likewise, the arrest rates for those values omit the portion of the corresponding arrest rates due to arrests for robbery and burglary. While not reported, the coefficients for the robbery and burglary rates were extremely statistically significant and positive. Entire sample used over the 1977 to 1992 period. *The result is statistically significant at the 1 percent level for a two-tailed t-test.
CONCEALED-HANDGUN LAWS AND CRIME RATES/73
was rising right up until the law was adopted but falling thereafter, some values that appeared while crime rate was rising could equal some that appeared as it was falling. In other words, deceptively similar levels can represent dramatically different trends over time.
I used several methods to examine changes in the trends exhibited over time in crime rates. First, I reestimated the regressions in table 4.1, using year-to-year changes on all explanatory variables (see table 4.6). These regressions were run using both a variable that equals 1 when a nondiscretionary law is in effect as well as the change in that variable (called "differencing" the variable) to see if the initial passage of the law had an impact. The results consistently indicate that the law lowered the rates of violent crime, rape, and aggravated assault. Nondiscretionary laws discourage murder in both specifications, but the effect is only statistically significant when the nondiscretionary variable is also differenced. The property-crime results are in line with those of earlier tables, showing that nondiscretionary laws produce increases in property crime. Violent crimes decreased by an average of about 2 percent annually, whereas property crimes increased by an average of about 5 percent.
As one might expect, the nondiscretionary laws affected crime immediately, with an additional change spread out over time^Why would the entire effect not be immediate? An obvious explanation is that not everyone who would eventually obtain a permit to carry a concealed handgun did so right away. For instance, as shown by the data in table 4.7, the number of permits granted in Florida, Oregon, and Pennsylvania was still increasing substantially long after the nondiscretionary law was put into effect. Florida's law was passed in 1987, Oregon's in 1990, and Pennsylvania's in 1989.
Reestimating the regression results from table 4.1 to account for different time trends in the crime rates before and after the passage of the law provides consistent strong evidence that the deterrent impact of concealed handguns increases with time. For most violent crimes, the time trend prior to the passage of the law indicates that crime was rising. The results using the simple time trends for these violent-crime categories are reported in table 4.8. Figures 4.5 through 4.9 illustrate how the violent-crime rate varies before and after the implementation of nondiscretionary concealed-handgun laws when both the linear and squared time trends are employed. Comparing the slopes of the crime trends before and after the enactment of the laws shows that the trends become more negative to a degree that is statistically significant after the laws were passed. 36
These results answer another possible objection: whether the findings are simply a result of so-called crime cycles. Crime rates rise or fall over
Table 4.6 Results of rerunning the regressions on differences
Endogenous variables in terms of first differences of the natural logarithm of the crime rate
Exogenous variables
Aln(violent- Aln(Murder Aln(Rape Aln(Aggravated-crime rate) rate) rate) assault rate)
Aln(robbery Aln(property- Aln(Burglary Aln(Larceny Aln(Auto-rate) crime rate) rate) rate) theft rate)
All variables except for the nondiscretionary dummy differenced
Nondiscretionary law adopted
First differences in the arrest rate for the crime category
First differences in the dummy for nondiscretionary law adopted
First differences in the arrest rate for the crime category
-22%*
-0.05%*
-2.6%
-0.15%*
-5.2%* -4.6%*
-0.09%*
-0.09%*
-3.3%*
5.2%*
-0.0
3.5%*
-0.24%*
All variables differenced
-2.7%*
-0.05%*
-3.6%*
-0.15%*
-3.9%*
-5.4%*
-0.09%* -0.09%*
-0.7%
-0.06%*
-0.0
0.7%
-.24%*
5.2%*
-0.02%*
6.2%*
-0.02%*
12.8%*
-0.02%*
24.2%*
-0.02%*
Note: The variables for income; population; race, sex, and age of the population; and density are all in terms of first differences. While not all the coefficient estimates are reported, all the control
variables used in Table 4.1 are used here, including year and county dummies. All regressions use weighted least squares, where the weighting is each county's population. Entire sample used over
the 1977 to 1992 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 result is statistically significant at the 11 percent level for a two-tailed t-test.
CONCEALED-HANDGUN LAWS AND CRIME RATES/75
Table 4.7 Permits granted by state: Florida, Oregon, and Pennsylvania
"Estimate of the number of concealed-handgun permits issued immediately before Florida's law
went into effect from David McDowall, Colin Loftin, and Brian Wiersema, "Easing Concealed
Firearms Laws: Effects on Homicide in Three States," Journal of Criminal Law and Criminology, 86 (Fall
1995): 194.
December 31, 1991.
'Number of permits issued under discretionary law.
time. If concealed-handgun laws were adopted at the peaks of these cycles (say, because concern over crime is great), the ensuing decline in crime might have occurred anyway without any help from the new laws. To deal with this, I controlled not only for national crime patterns but also for individual county patterns by employing burglary or robbery rates to explain the movement in the other crime rates. I even tried to control for individual state trends. Yet the simplest way of concisely illustrating that my results are not merely a product of the "normal" ups and downs in crime rates is to look again at the graphs in figures 4.5—4.9. With the exception of aggravated assault, the drops not only begin right when the laws pass but also take the crime rates well below what they had been before the passage of the laws. It is difficult to believe that, on the average, state legislatures could have timed the passage of these laws so accurately as to coincide with the peaks of crime waves; nor can the resulting declines be explained simply as reversions to normal levels.
Was the Impact of Nondiscretionary Concealed-Handgun Laws the Same Everywhere?
Just as we found that the impact of nondiscretionary laws changed over time, we expect to find differences across states. The reason is the same in both cases: deterrence increases with the number of permits. While the information obtained from state government officials only pertained to why permits were issued at different rates across counties within a
Toble 4.8 Chonge in time trends for crime rotes before ond after the adoption of nondiscretionory lows
Percent change in various crime rates for change in explanatory variable
Violent Aggravated
crime Murder Rape assault
Property Robbery crime Auto theft Burglary Larceny
Change in the crime rate from the difference in the annual change in crime rates in the years before and after the change in the law (annual rate after the law — annual rate before the law)
-0.9%*
-356*
-1.456* -0.5%*
-2.7%*
-0.6%*
-0.3%*
-1.5%*
-0.1%
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. Entire sample used over the 1977 to 1992
period.
*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.
-4-2024
Years before and after the adoption
of concealed-handgun laws
Figure 4.5. The effect of concealed-handgun laws on violent crimes
-6-4-2 0 2 4
Years before and after the adoption of concealed-handgun laws
Figure 4.6. The effect of concealed-handgun laws on murders
3 Q. O
Q.
E
-10
10
Years before and after the adoption of concealed-handgun laws
Figure 4.7. The effect of concealed-handgun laws on rapes
a o
Q.
.Q 9
E
-10
10
Years before and after the adoption of concealed-handgun laws
Figure 4.8. The effect of concealed-handgun laws on robbery rates
CONCEALED-HANDGUN LAWS AND CRIME RATES/79
Years before and after the adoption of concealed-handgun laws
Figure 4.9. The effect of concealed-handgun laws on aggravated assaults
given state, the rate at which new permits are issued at the state level may also vary based upon population and population density. If this is true, then it should be possible to explain the differential effect that non-discretionary laws have on crime in each of the states that passed such laws in the same way that we examined differences across counties.
Table 4.9 reexamines my earlier regressions, where I took into account that concealed-handgun laws have different effects across counties, depending upon how lenient officials had been in issuing permits under a previously discretionary system. The one change from earlier tables is that a different coefficient is used for the counties in each of the ten states that changed their laws during the 1977 to 1992 period. At least for violent crimes, the results indicate a very consistent effect of nondiscretionary concealed-handgun laws across states. Nine of the ten states experienced declines in violent-crime rates as a result of these laws, and eight of the ten states experienced declines in murder rates; in the states where violent crimes, murders, or robberies rose, the increases were very small. In fact, the largest increases were smaller than the smallest declines in the states where those crime rates fell.
Generally, the states with the largest decreases in any one category tended to have relatively large decreases across all the violent-crime categories, although the "leader" in each category varied across all the
Table 4.9 State-specific impact of nondiscretionary concealed-handgun laws
Note: The table uses arrest rates adjusted for counties wherein the adoption of nondiscretionary concealed-handgun laws was most likely to represent a real change from past practice by multiplying the nondiscretionary-law variables by the population in each county. The percents are evaluated at the mean county population.
CONCEALED-HANDGUN LAWS AND CRIME RATES/81
violent-crime categories. 37 Likewise, the states with relatively small crime decreases (for example, Georgia, Oregon, Pennsylvania, and Virginia) tended to exhibit little change across all the categories.
Property crimes, on the other hand, exhibited no clear pattern. Property crimes fell in five states and increased in five states, and the size of any decrease or increase was quite small and unsystematic.
Ideally, any comparison across states would be based on changes in the number of permits issued rather than simply the enactment of the
nondiscretionary law States with the largest increases in permits should show the largest decreases in crime rates. Unfortunately, only a few states have recorded time-series data on the number of permits issued. I will use such data in chapter 5. For the moment, it is still useful to see whether the patterns in crime-rate changes found earlier across counties are also found across states. In particular, we would like to know whether the largest declines occurred in states with the largest or most dense populations, which we believed had the greatest increase in permits. The justification for the county-level differences was very strong because it was based on conversations with individual state officials, but those officials were not asked to make judgments across states (nor was it likely that they could do so). Further, there is much more heterogeneity across counties, and a greater number of observations. The relationship posited earlier for county populations also seems particularly tenuous when dealing with state-level data because a state with a large population could be made up of a large number of counties with small populations.
With this list of reservations in mind, let us look at the results we get by using state-level density data. Table 4.10 provides the results with respect to population density, and we find that, just as in the case of counties, larger declines in crime were recorded in the most densely populated states. The differences are quite large: the most densely populated states experienced decreases in violent crimes that were about three times greater than the decreases in states with the average density. The results were similar when state populations were taken into account.
Other Gun-Control Laws and Different Types of Concealed-Handgun Laws
Two common restrictions on handguns arise from (1) increased sentencing penalties for crimes involving the use of a gun and (2) waiting periods required before a citizen can obtain a permit for a gun. How did these two types of laws affect crime rates? Could it be that these laws—rather than concealed-handgun laws—explain the deterrent effects? To answer this question, I reestimated the regressions in tables 4.1 and 4.3 by
Table 4.10 Effects of concealed-handgun lows across states related to differences in state population density
Note: The regressions used for this table multiplied the variable for whether the law was enacted by that state's population density. The control variables used to generate these estimates 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 state's population.