Appendix A
Dissent
James Q. Wilson
The thrust of Chapter 6 of the committee’s report is that studies purporting to show a relationship between right-to-carry (RTC) laws and crime rates are fragile. Though I am not an econometrician, I am struck by the fact that most studies of the effect of policy changes on crime rates are fragile in this sense: Different authors produce different results, and sometimes contradictory ones. This has been true of studies of the effect on crime rates of incapacitation (that is, taking criminals off the street), deterrence (that is, increasing the likelihood of conviction and imprisonment), and capital punishment. In my view, committees of the National Research Council that have dealt with these earlier studies have attempted, not simply to show that different authors have reached different conclusions, but to suggest which lines of inquiry, including data and models, are most likely to produce more robust results.
That has not happened here. Chapter 6 seeks to show that fragile results exist but not to indicate what research strategies might improve our understanding of the effects, if any, of RTC laws. To do the latter would require the committee to analyze carefully not only the studies by John Lott but those done by both his supporters and his critics. Here, only the work by Lott and his coauthors is subject to close analysis.
If this analysis of Lott’s work showed that his findings are not supported by his data and models, then the conclusion that his results are fragile might be sufficient. But my reading of this chapter suggests that some of his results survive virtually every reanalysis done by the committee.
Lott argued that murder rates decline after the adoption of RTC laws even after allowing for the effect of other variables that affect crime rates.
The committee has confirmed this finding as is evident in its Tables 6-1, 6-2, 6-5 (first row), 6-6 (first row), and 6-7 (first two rows). This confirmation includes both the original data period (1977-1992) used by Lott and data that run through 2000. In view of the confirmation of the findings that shall-issue laws drive down the murder rate, it is hard for me to understand why these claims are called “fragile.”
The only exceptions to this confirmation are, to me, quite puzzling. Tables 6-5 and 6-6 suggest that RTC laws have no effect on murder rates when no control variables are entered into the equations. These control variables (which include all of the social, demographic, and public policies other than RTC laws that might affect crime rates) are essential to understanding crime. Suppose Professor Jones wrote a paper saying that increasing the number of police in a city reduced the crime rate and Professor Smith wrote a rival paper saying that cities with few police officers have low crime rates. Suppose that neither Jones nor Smith used any control variables, such as income, unemployment, population density, or the frequency with which offenders are sent to prison in reaching their conclusions. If such papers were published, they would be rejected out of hand by the committee for the obvious reason that they failed to supply a complete account of the factors that affect the crime rate. One cannot explain crime rates just by observing the number of police in a city any more than one can explain them just by noting the existence of RTC laws.
It is not enough to say that it is hard to know the right set of control variables without calling into question the use of economics in analyzing public policy questions. All control variables are based on past studies and reasonable theories; any given selection is best evaluated by testing various controls in one’s equations.
In addition, with only a few exceptions, the studies cited in Chapter 6, including those by Lott’s critics, do not show that the passage of RTC laws drives the crime rates up (as might be the case if one supposed that newly armed people went about looking for someone to shoot). The direct evidence that such shooting sprees occur is nonexistent. The indirect evidence, as found in papers by Black and Nagin and Ayres and Donohue [cited in Chapter 6], is controversial. Indeed, the Ayres and Donohue paper shows that there was a “statistically significant downward shift in the trend” of the murder rate (Chapter 6, page 135). This suggests to me that for people interested in RTC laws, the best evidence we have is that they impose no costs but may confer benefits. That conclusion might be very useful to authorities who contemplate the enactment of RTC laws.
Finally, the committee suggests that extending the Lott model to include data through 2000 may show no effect on RTC laws on murder rates if one analyzes the data on a year-by-year basis (Table 6-7, rows three and four). I wish I knew enough econometrics to feel confident about this
argument, but I confess that at first blush it strikes me as implausible. To me, Lott’s general argument is supported even though it is hard to assign its effect to a particular year. Estimating the effects of RTC laws by individual years reduces the number of observations and thus the likelihood of finding a statistically significant effect. It is possible that doing this is proper, but it strikes me that such an argument ought first to be tested in a peer-reviewed journal before it is used in this report as a sound strategy.
Even if the use of newer data calls into question the original Lott findings, a more reasonable conclusion is that Lott’s findings depend on crime rate trends. The committee correctly notes that between 1977 and 1992 crime rates were rising rapidly while between 1993 and 1997 they were declining. Lott’s original study was of the first time period. Suppose that his results are not as robust for the second period. The committee concludes that this shows that his model suffers from “specification errors” (page 141). Another and to me more plausible conclusion is that the effect of RTC laws on some crime rates is likely to be greater when those rates are rising than when they are falling. When crime rates are rising, public policy interventions (including deterrence, incapacitation, and RTC laws) are likely to make a difference because they create obstacles to the market and cultural forces that are driving crime rates up. But when crime rates are falling, such interventions may make less of a difference because they will be overwhelmed by market and cultural changes that make crime less attractive. This may or may not be a reasonable inference, but it is worthy of examination.
In sum, I find that the evidence presented by Lott and his supporters suggests that RTC laws do in fact help drive down the murder rate, though their effect on other crimes is ambiguous.