More Guns Less Crime
Page 20
One of the commentators on my book suggested that in addition to year-to-year changes in the national crime rate as well as state and county crime trends, another way to account for crime cycles is by measuring whether the crime rates are falling faster in right-to-carry states
than in other states in their region rather than compared to just the nation as a whole. While it is impossible to use a separate variable for each year for each individual state, because that would falsely appear to explain all the year-to-year changes in average crime rates in a state, it is possible to group states together. This new set of estimates would account not only for whether the crime rates in concealed-handgun states are falling relative to the national crime rate but now also for whether they are falling relative to the crime rates in their region. To do this, the country is divided into five regions (Northeast, South, Midwest, Rocky Mountains, and Pacific) and variables are added to measure the year-to-year changes in crime by region. n All county- and city-level regressions will employ these additional control variables.
Some have criticized my earlier work for not doing enough to account for poverty rates. As a response, I have incorporated in this section of the book state-level measures of poverty and unemployment rates in addition to all the county-level variables that accounted for these factors earlier in this book. The execution rates for murders in each state are now included in estimates to explain the murder rate. Finally, new data on the number of permits granted in different states make it easier to link crime rates to the number of permits granted.
Reviewing the Basic Results
The central question is, How did crime rates change before and after the right-to-carry laws went into effect? The test used earlier in this book examined the difference in the time trends before and after the laws were enacted. 12 With the extended data and the additional variables for the year-to-year changes in crime by region (so-called regional fixed year effects), state poverty, unemployment, and death-penalty execution rates, table 9.1 shows that this pattern closely resembles the pattern found earlier in the book: violent-crime rates were rising consistently before the right-to-carry laws and falling thereafter. 13 The change in these before-and-after trends was always extremely significant—at least at the 0.1 percent level. Compared to the results for tables 4.8 or 4.13, the effects were larger for overall violent crimes, rape, robbery, and aggravated assaults and smaller for murder. For each additional year that the laws were in effect, murders fell by an additional 1.5 percent, while rape, robbery, and aggravated assaults all fell by about by 3 percent each year. The other variables continued to produce results similar to those that were found earlier. 14
While no previous crime study accounts for year-to-year changes in regional crime rates, it is possible to go even beyond that and combine
Table 9.1 Reexamining the change in time trends before and after the adoption of nondiscretionary laws, using additional data for 1995 and 1996
Percent change in various crime rates for changes in explanatory variables
Violent Aggravated Property
crime Murder Rape Robbery assault crime
Burglary Larceny
Auto theft
Change in the crime rate from the difference in the annual change in crime rates in the years before and after the adoption of the right-to-carry law (annual rate of change after the law — annual rate of change before the law)
-13%*
-1.5%*
-3.2%*
-1.9%*
-3.0%*
-1.6%*
-2.5%*
-0.9%*
-2.1%*
Note: This table uses county-level violent and property-crime data from the Uniform Crime Report that were not available when I originally wrote the book. All regressions use weighted least squares, where the weighting is each county's population. The regressions correspond to those in tables 4.8 and 4.13. The one difference from the earlier estimates is that these regressions now also allow the regional fixed effects to vary by year.
different approaches. Including not only the factors accounted for in table 9.1 but also individual state time trends produces similar results. The annual declines in crime from right-to-carry laws are greater for murder (2.2 percent), rape (3.9 percent), and robbery rates (4.9 percent), while the impact on aggravated assaults (0.8 percent) and the property crime rates (0.9 percent) is smaller.
Figures 9.1-9.5 illustrate how the violent-crime rates vary before and
400-
300"
-10 -8 -6 -4 -2 0 2 4 6 8 10 Years before and after the adoption of concealed-handgun laws Figure 9.1. The effect of concealed-handgun laws on violent crimes
-10 -6 0 5 10
Years before and after the adoption of concealed-handgun laws
Figure 9.2. The effect of concealed-handgun laws on murders
8
-10-5 0 5 10
Years before and after the adoption of concealed-handgun laws Figure 9.3. The effect of concealed-handgun laws on rapes
140
I
-10 -5 0 5 10
Years before and after the adoption of concealed-handgun laws Figure 9.4. The effect of concealed-handgun laws on robberies
after the implementation of right-to-carry laws when both the linear and squared time trends are employed. Despite expanding the data through 1996 so that the legal changes in ten additional states could be examined, the results are similar to those previously shown in figures 4.5—4.9. 15 As in the earlier results, the longer the laws are in effect, the larger the decline in violent crime. The most dramatic results are again for rape and
9 "5
-10-5 0 5 10
Years before and after the adoption of concealed-handgun laws Figure 9.5. The effect of concealed-handgun laws on aggravated assaults
robbery rates, which were rising before the right-to-carry law was passed and falling thereafter. Robbery rates continue rising during the first full year that the law is in effect, but the rate of increase slows and begins to fall by the second year. It is this continued increase in robbery rates which keeps the violent crimes as a whole from immediately declining. While aggravated assaults were falling on average before the right-to-carry law was adopted, figure 9.5 shows that the rate of decline accelerated after the law went into effect.
What Determines the Number of Permits Issued and What Is the Net Benefit from Issuing Another Permit?
The Number of Permits
The relationship between the percentage of the population with permits and the changes in crime rates is central to much of the debate over the right to carry. My previous work was based on the number of permits issued for counties in Oregon and Pennsylvania as well as on discussions with various government officials on what types of counties issued the most permits. The comparison across states assumed that what created the difference in permit rates across counties also applied across states. Some more state-level data have now become available on permit rates, but such data are still relatively scarce. In addition to Florida, Oregon,
and Pennsylvania, I have also acquired some annual permit-rate data up to 1996 for Alaska, Arizona, Oklahoma, South Carolina, Texas, Utah, and Wyoming, though these states had these rules in effect for no more than a few years.
While these data are limited, they allow us to examine what factors determine permitting rates, which in turn lets us link the permitting rate to changes in crime. Permit prices, the amount of training required to get a permit, the length of time that permitting rules have been in effect, and the crime rate are all important factors in determining how many people will get permits. Permitting fees and prices charged for training courses are expected to reduce the number of permits issued, but another important cost of getting a permit is the time spent meeting the requirements. This is not to say that there are not also benefits from training (that is a separate issue), but in the narrow issue of how many permits will be issued, there is no doubt that longer training requirements discourage some people fr
om getting permits.
What permitting rules are in place largely depends upon when the laws were first enacted. States that adopted right-to-carry laws more recently tend to have more restrictive licensing requirements. For example, the three states (Alaska, Arizona, and Texas) requiring at least ten hours of training adopted their rules during the last few years of the sample, and Arizona is the only right-to-carry state that requires additional training when permits are renewed. Six of the eight states with permitting fees of at least $100 have also enacted the law during the last few years. This raises the concern that the drops in crime from the passage of right-to-carry laws may be smaller in the states that have most recently adopted these laws simply because they have issued fewer permits.
Based on state-level data, table 9.2 shows the impact of permit fees, training requirements, and how long (in years) the law has been in effect. Because the evidence indicates that the number of new permits is likely to trail off over time, the estimates include both the number of years the law has been in effect and the number of years squared. Fees and training requirements were first investigated without square terms. Notice that only a small fraction of the population gets permits, ranging from less than 1 percent to 6 percent. With that in mind, the regression results show that for each $10 increase in fees, the population getting permits is reduced by about one half of a percentage point. And requiring five hours of training (rather than none) reduces the number of permits by about two-thirds of a percentage point. In a typical state without any fees or training requirements, the percentage of the population with permits would grow from about 3 percent to a little less than 6 percent after a decade.
Table 9.2 What determines the rate at which people obtain permits?
Percentage of the state population with permits
-.5%*
6.1%*
*The result is significant at the 1 percent level for a two-tailed t-test.
I also ran more complicated specifications including squared terms for fees and training requirements. They give similar results: fees discourage people from obtaining permits over almost the entire range (until fees go over $130, which is near the highest fee in the sample—$140 for Texas). Anecdotal evidence from newspapers indicates that yet another factor is important: the fear of an attack. Thus, crime and multiple-victim public shootings increase gun sales and concealed-handgun permits. 16 Other variables, such as violent-crime rates, murder rates, the number of multiple-victim public shootings, or the death rate from those attacks, are also important for determining how many people get permits, but they do not alter the impact of the previously mentioned variables. Each additional multiple-victim public shooting increases a state's number of permits by about two-tenths of a percentage point, and each additional person who is killed in such a shooting (per 1 million people living in a state) increases handgun permits by one-tenth of a percentage point.
The Crime Rate and the Estimated Number of Concealed Handguns
The above estimates allow us to revisit the impact of permits and crime rates. While the time-series data on permits issued in different states are relatively short, we do have detailed information on the factors that help determine the number of permits (the fees, training requirements, and how long the law has been in effect). The results from the specification shown in table 9.2 were used to construct "predicted values." Constructing a predicted percentage of a state's population with permits allows us to do more than relying on how crime rates change over time or on the anecdotal evidence I obtained from surveying different state permitting agencies.
These new results using state-level data, shown in table 9.3, indicate that violent-crime rates fell across the board as more permits were issued, with the largest drop occurring for robberies. These results correspond closely to the diagrams reported in figures 4.6—4.9 and 7.1—7.4, which indicate that robberies and rapes are most dramatically affected by the number of years that right-to-carry laws are in effect. The coefficients imply that for every 1,000 additional people with permits, there are 0.3 fewer murders, 2.4 fewer rapes, 21 fewer robberies, and 14.1 fewer aggravated assaults. 17 On the other hand, with the exception of burglary, property crime remained statistically unchanged as more people obtained permits.
Would society benefit from more people getting permits? As already noted, obtaining a permit costs money and takes time. Carrying around a gun is also inconvenient, and many states impose penalties if the gun does not remain concealed. 18 On the positive side, permit holders benefit from having the gun for protection and might also come to the rescue of others. But perhaps just as important are the benefits to general crime deterrence produced by concealed-carry laws, for they also help protect others indirectly, as criminals do not know which people can defend themselves until they attack. This raises the real risk that too few people will get permits, as permit holders personally bear all these costs but produce large benefits for others.
Whether too few permits are being issued depends on how the crime rate changes as more and more permits are issued and whether it is the permit holder or the general public who primarily reaps the benefit from more concealed carry.
The impact of increasing the number of permits on crime is shown in table 9.3, column 1. However, the impact does not need to be constant as more people get permits. Indeed, there may well exist what economists call "diminishing returns"—that is, the crime-reducing benefits from another person getting a permit falls as more people get permits. The reason behind this is twofold: first, those most at risk could be the first to get permits; second, once one adult in a public setting (e.g., a store) has a concealed handgun, the additional benefit from a second or third person being armed should be relatively smaller.
But it is also conceivable that the probability that a victim can defend herself must rise above a certain threshold before it does much to discourage criminals. For instance, if only a few women brandish guns, a would-be rapist may believe that a defensive use is simply an exception and go after another woman. Perhaps if a large enough percentage of women defend themselves, the would-be rapist would decide that the risk to himself is too high.
Table 9.3 Using the predicted percent of the population with permits to explain the changes in different crime rates for state data
One-percentage-point change in the share of the state population with permits
(i)
Pattern when a quadratic term is added for the percent of the population with permits (2)
Number by which total crimes are reduced when an additional 1 percent of the population obtains permits in 1996, using the estimates from column 1 for states that had a right-to-carry law in effect by that year (3)
Violent crime
Robbery
-7%*
Murder -4%***
Rape -7%*
-13.6%*
Drop reaches its maximum when
23% of the population has
permits
Drop reaches its maximum when
&% of the population has permits
Drop is increasing at an
increasing rate as more people
get permits
Drop tapers off, but so slowly
that it is still falling when 100
percent of the population has
permits
432 lives saved 3,862 fewer rapes
35,014 fewer robberies
Aggravated assault
Property
crime
Burglary
Larceny Auto theft
-5%**
-2.6%*** -10%*
-.6% -3%
Drop reaches its maximum when
6 percent of the population has
permits
Drop continues at a constant rate
Drop is increasing at an
increasing rate as more people
get permits
No significant pattern
Drop reaches its maximum when
3 percent of the population
has
permits
28,562 fewer aggravated assaults
144,227 fewer burglaries
27,922 fewer larcenies 21,254 fewer auto thefts
Note: Using the National Institute of Justice estimates of what crime costs victims to estimate the net savings from 1 percent more of the population obtaining permits (or of each additional permit) in 1998 dollars, the cost is reduced by $3.45 billion ($2,516 per permit). *The result is significant at the 1 percent level for a two-tailed t-test. **The result is significant at the 5 percent level for a two-tailed t-test. ***The result is significant at the 15 percent level for a two-tailed t-test.
One can test for diminishing returns from more permits by using a squared term for the percentage of the population with permits. The results (shown in column 2) indicate that right-to-carry states experience additional drops in all the violent-crime categories when more permits are issued. For murder, rape, and robbery, all states experience further reductions in crime from issuing more permits, though diminishing returns appear for murder and aggravated assault. (Only one state—Pennsylvania—approaches the number of permits beyond which there would be little further reduction in aggravated assaults from issuing more permits.) An important word of caution is in order here. These particular estimates of the percentage of the population that minimizes crime are rather speculative, because they represent predictions outside the range for which observed permit levels are available. (We thus cannot use these results to predict with confidence what would happen if a state got up to, say, 8 percent having permits.) Still, there is little doubt that issuing additional permits beyond what we have today lowers crime.