More Guns Less Crime (30 page)

Figures 9.10—9.13 present the range of estimates associated with these different combinations of variables and specifications, both in terms of their extreme bounds and their median value. What immediately stands out when one examines all these estimates is how extremely consistent the violent-crime results are. For example, take figure 9.10. A one-percentage-point change in people with permits lowers violent-crime rates by 4.5—7.2 percent. Indeed, all the estimates (over two thousand of them) for overall violent crime, murder, rape, robbery, and aggravated assault indicate that increases in permits reduce crime. All the combinations of the other ten sets of control variables imply that a one-percentage-point increase in the population holding permits reduces murder rates by 2-3.9 percent annually. Compared to the state-level data, the benefits from right-to-carry laws are much smaller for robbery and much larger for aggravated assaults.

Figure 9.11 uses the simple before-and-after trends to examine the impact of the right-to-carry laws, and the results for the violent-crime rates

it

I 1

§§

$2.

•§*

II

1!

15

Property

EPILOGUE/ 187

Larceny

tl

Robbery

Violent crime

Rape

Aggravated

Crime categories

Figure 9.10. Sensitivity of the relationship between the percentage of the population with permits and annual changes in crime rates: data for all counties

1

£ -0.005

a> -0

2 -0.015"

a

8

■g -0.03

I

8. -0.035-■

09

5

Larceny

Auto theft

Violent crime

Robbery

♦Largest relative increase in crime (smallest relative drop)

■ Median effect

ALargest relative drop in crime (smallest relative increase)

Burglary

Aggravated

Rape

Crime categories

Figure 9.11. Sensitivity of the relationship between right-to-carry laws and annual changes in crime rates: data for all counties

are generally consistent with those shown in figure 9.10. Again, all the violent-crime-rate regressions show the same direction of impact from the concealed-handgun law. The median estimated declines in violent-crime rates are quite similar to those initially reported in table 9. 1. For each additional year that the right-to-carry laws are in effect, violent

6 4 2 •'

Property crime

Larceny

r

Robbery

Burglary

Violent

w n crime for

crime'for counties

Rape

Violent

counties

with more

with more *»>

than 2 °. 000

100,000 PeoP' 6 people

Aggravated assault

♦Largest relative increase in crime (smallest relative drop)

■ Median effect

▲Largest relative drop in crime (smallest relative increase)

Figure 9.12. Sensitivity of the relationship between the percentage of the population with permits and annual changes in crime rates: data for counties with either more than 20,000 people or more than 100,000 people (all individual crime categories—that is, all categories except "violent crime"—are for counties with more than 20,000 people)

1

•g> -0.01

I 002 1

■§ -0.03-

8. -0.035-

1

c

< -0.04-

Larceny

Property crime

Auto theft

Burglary

Violent Violent

crime for crime for

counties counties

with more with more

than than

100,000 20,000

Robbery

people

people

Rape

Aggravated assault Crime categories

♦Largest relative increase in crime (smallest relative drop)

effect ▲Largest relative drop in crime (smallest relative increase)

Figure 9.13. Sensitivity of the relationship between right-to-carry laws and annual changes in crime rates: data for counties with either more than 20,000 people or more than 100,000 people (all individual crime categories—that is, all categories except "violent crime"—are for counties with more than 20,000 people)

crimes decline by 2.4 percent, murders by 1.6 percent, rapes and aggravated assaults by over 3 percent, and robberies by 2.7 percent.

With the notable exception of burglaries, which consistently decline, figures 9.10 and 9.11 provide mixed evidence for whether xight-to-carry laws increase or decrease other property crimes. Even when one focuses on estimates of one type, such as those using the percentage of the population with permits, the county- and state-level data yield inconsistent results. Yet while the net effect of right-to-carry laws on larceny and auto theft is not clear, one conclusion can be drawn: the passage of right-to-carry laws has a consistently larger deterrent effect against violent crimes than property crimes and may even be associated with increases in property crimes.

Figures 9.12 and 9.13 limit the sample to the more populous counties and continue reaching very similar results. For counties with more than 20,000 people, the estimate ranges are always of the same sign and have magnitudes similar to those results which examined all the counties. Both figures also looked at the sensitivity of the overall violent-crime rate for counties over 100,000. The range of estimates was again very similar, though they implied a slightly larger benefit than for the more populous counties. For example, figure 9.12 shows that in counties with more than 20,000 people violent crime declines by between 5.4 and 7.4 percentage points for each additional 1 percent of the population with permits, while the analogous drop for counties with more than 100,000 people is between 5.8 and 8.7 percentage points.

A total of 13,312 regressions for the various violent-crime categories are reported in this section. The evidence clearly indicates that right-to-carry laws are always associated with reductions in violent crime, and 89 percent of the results are statistically significant at least at the 1 percent level. The results are not sensitive to including particular control variables and always show that the benefits from these laws increase over time as more people obtain permits. The 8,192 regressions for property crime imply a less consistent relationship between right-to-carry laws and property crime, but even when drops in property crime are observed, the declines are smaller than the decrease in violent crime.

While limiting the sample size to only larger-population counties provides one possible method of dealing with "undefined" arrest rates, it has a serious drawback—information is lost by throwing out those counties with fewer than 20,000 people. Another approach is to control for either the violent- or property-crime arrest rate depending upon whether the crime rate being studied is that of violent or property crime. Even if a county has zero murders or rapes in a particular year, virtually all counties have at least some violent or property crime, thus eliminating the

"undefined" arrest rate problem and still allowing us to account for county-level changes over time in the effectiveness of law enforcement. This approach also helps mitigate any spurious relationship between crime and arrest rates that might arise because the arrest rate is a function of the crime rate. Reestimating the 4,096 regressions in figure 9.10 for murder, rape, robbery, aggravated assault, auto theft, burglary, and larceny with this new measure of arrest rates again produces very similar results.

City Crime Data

County data, rather than city data, allow the entire country to be examined. This is important, since, obviously, not everyone lives in cities. Such data further allow us to deal with differences in how permits are issued, such as the discretion states grant to local law enforcement. Relying on county data allows a detailed analysis of many important factors, such as arrest and conviction rates, the number of police, expenditures on police, (sometimes) prison sentences, and proxies for policing policies like the so-called broken-windows strategy (according to which police focus on less serious property crimes as a means of reducing overall violent crime). Yet a drawback with county data is that policing policies cannot be dealt with well, for such policy decisions are made at the level of individual police departments—not at the county level. 27 With a few exceptions such as San Francisco, Philadelphia, and New York, where county and city boundaries coincide, only city-level data can be used to study these issues.

The focus of my research is guns and crime, but I had to make sure that I accounted for whatever policing policies are being employed. 28 Three policing strategies dominate the discussion: community-oriented policing, problem-oriented policing, and the broken-windows approach. While community-oriented policing is said to involve local community organizations directly in the policing effort, problem-oriented policing is sometimes viewed as a less intrusive version of the broken-windows policy. Problem-oriented policing began as directing patrols on the basis of identified crime patterns but nowadays involves the police in everything from cleaning housing projects and surveying their tenants to helping citizens design parking garages to reduce auto theft. 29 An extensive West-law database search was conducted to categorize which cities adopted which policing strategies as well as their adoption and rescission dates. 30

Other recent research of mine demonstrates the importance of racial and gender hiring decrees on the effectiveness of police departments. 31 When hiring rules are changed so as to create equal pass rates on hiring

EPILOGUE/ 191

exams across different racial groups—typically by replacing intelligence tests with what some claim are arbitrary psychological tests—the evidence indicates that the quality of new hires falls across the board. And the longer these new hiring policies are in place, the more detrimental the effect on police departments. As with the right-to-carry laws, simple before-and-after trends were included to measure the changing impact of these rules over time.

Let us return to the main focus, guns and crime. To examine the impact of right-to-carry laws, the following list of variables has been accounted for: city population, arrest rate by type of crime, unemployment rate, percentage of families headed by females, family poverty rate, median family income, per-capita income, percentage of the population living below poverty, percentage of the population that is white, percentage that is black, percentage that is Hispanic, percentage that is female, percentage that is less than five years of age, percentage that is between five and seventeen, percentage that is between eighteen and twenty-five, percentage that is between twenty-six and sixty-four, percentage that is sixty-five and older, median population age, percentage of the population over age twenty-five with a high school diploma, percentage of the population over age twenty-five with a college degree, and other types of gun-control laws (waiting periods, background checks, and additional penalties for using guns in the commission of a crime). As with the earlier county- and state-level data, variables are included to measure the length of state waiting periods, as well as the change in average crime rates from state waiting periods, background checks, penalties for using a gun in the commission of crime, and whether the federal Brady law altered existing state rules. Again, all estimates include variables to account for the average differences across counties and years as well as by year within region.

Table 9.4 provides strong evidence that even when detailed information on policing policies is taken into account, passing concealed-handgun laws deters violent crime. The benefit in terms of reduced murder rates is particularly large, with a drop of 2.7 percent each additional year that the right-to-carry law is in effect. The drop experienced for rapes is 1.5 percent per year. The one violent crime for which the decline is not statistically significant is aggravated assault. On the other hand, property crimes increase after the adoption of right-to-carry laws, confirming some of the earlier findings.

Consent decrees—which mandate police hiring rules that ensure equal pass rates by race and sex—significantly and adversely affect all crime categories but rape. For each additional year that the consent decree is in effect, overall violent crimes rise by 2.4 percent and property crimes rise by 1.9 percent.

Toble 9.4 Accounting for policing policies using city-level doto

Percent change in various crime rates for changes in explanatory variables

Violent

crime Murder

Aggravated Property Auto

Rape Robbery assault crime Burglary Larceny theft

Change in the crime rate from —1.2%**

the difference in the annual

change in crime rates in the

years before and after the

adoption of a right-to-carry

law (annual rate of change

after the law — annual rate

of change before the law) Change in the crime rate after 2.4%*

imposition of a consent

decree regarding the hiring

of police officers Change in the average crime —3.3% a

rate after implementation of

community policing

-2.7%*

0.4%*

2.2%

-1.5%** -1.0%* -0.6% 0.92%** 1.0

-0.27% 3.5%* 1.4%*

1.9%*

0.7%** 1.2%*

2.4%* 2.0%* 0.5%*

rate after implementation of

problem-orientated policing Change in the average crime -0.8% 6.7% c -10.1% a -3.8% 2.3% -6.4% a -5.6% a -12.3% a 18.2% a

rate after implementation of

broken-window policing Average crime rate after 9.3% 14.7% d 6.8% 7.9% d 15.8% c -0.6% 2.7% -4.9% 11%

adoption of one-gun-a-

month purchase rule Change in the average crime 9.6% a 18.4% a 11.9% 4.1% 17.2% a 13% a 10.6% b 14.3% c 11% C

rate in a state after a