Abstract

In their reply to our comment on their initial paper, Moody and Marvell continue their analysis of right-to-carry (RTC) laws using panel data for the period 1977-2000. But with six additional years of data now available for analysis, we think the need for further parsing of older data is of limited value in assessing the more guns, less crime hypothesis. In this comment, we add six years of data to what Moody and Marvell previously analyzed. We show that, whether one looks at the original Lott and Mustard specification, the latest Moody and Marvell specification, or a plausible alternative specification, there is consistent evidence for the unsurprising proposition that RTC laws increase aggravated assault. We address some anomalies in these models and their resulting estimates. The Lott and Mustard model, for example, suffers from omitted-variable bias in failing to control for the impact of incarceration. In addition, the Moody and Marvell model generates odd predictions of the impact of incarceration on crime for most crime categories, and it appears to suggest (anomalously) that crack had no impact on murder. These and other problems raise questions about how well these regressions work to reveal the true effect on crime of RTC laws. For instance, would better data and models reveal that the estimated increases in murder and robbery are also statistically significant, as they are for the related violent crime of aggravated assault? Or might the estimated effect of aggravated assault be altered if we knew the full impact of changing police responses to domestic violence?