Economics, Law and Policy ISSN 2576-2060 (Print) ISSN 2576-2052 (Online) Vol. 6, No. 3, 2023 www.scholink.org/ojs/index.php/elp 13 Original Paper Does the Federal Law Forbidding People under Domestic Violence Restraining Orders from Possessing Firearms Save Lives? John R. Lott, Jr. 1 & Carlisle E. Moody 2* 1 Crime Prevention Research Center, Missoula, MT, USA 2 Department of Economics, College of William and Mary, Williamsburg, VA, USA * Carlisle E. Moody, Department of Economics, College of William and Mary, Williamsburg, VA, USA Received: September 15, 2023 Accepted: September 25, 2023 Online Published: October 18, 2023 doi:10.22158/elp.v6n3p13 URL: http://dx.doi.org/10.22158/elp.v6n3p13 Abstract The Supreme Court of the United States will hear arguments concerning Federal Law 18 U.S.C. 922(g)(8) which forbids individuals who are under a Domestic Violence Protection Order (DVPO) from possessing firearms. This paper analyzes the potential costs of overturning that law. We estimate a variety of models to determine the effect of the law on domestic murders, domestic femicides, domestic gun murders, and domestic gun femicides. We subject the analysis to a variety of robustness checks. The results are remarkably robust. We find that 18 U.S.C. § 922(g)(8) does not significantly reduce domestic murder, domestic femicide, domestic gun murder, or domestic gun femicide. Keywords domestic violence, firearm policy analysis, two-way fixed-effects, 18 U.S.C. 922(g)(8) www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 14 Published by SCHOLINK INC. 1. Introduction On February 2, 2023, the Fifth Circuit Court of Appeals vacated the conviction of Zackey Rahimi. A Federal court convicted Mr. Rahimi possessing firearms while under a Domestic Violence Protection Order (DVPO), violating section 18 U.S.C. 922(g)(8) of the US Code (hereafter “922” law), a law passed by Congress in 1994. The court concluded that, Doubtless, 18 U.S.C. § 922(g)(8) embodies salutary policy goals meant to protect vulnerable people in our society. Weighing those policy goals’ merits through the sort of means-end scrutiny our prior precedent indulged, we previously concluded that the societal benefits of § 922(g)(8) outweighed its burden on Rahimi’s Second Amendment rights. But Bruen forecloses any such analysis in favor of a historical analogical inquiry into the scope of the allowable burden on the Second Amendment right. Through that lens, we conclude that § 922(g)(8)’s ban on possession of firearms is an “outlier that our ancestors would never have accepted.” Id. Therefore, the statute is unconstitutional, and Rahimi’s conviction under that statute must be vacated (Note 1). Rahimi had a lengthy criminal history. In 2019, Rahimi fired a gun at a passerby who witnessed him dragging his girlfriend through a parking lot. Months later, he shot at a driver he had gotten into an accident with. And in 2020, he assaulted another girlfriend and threatened her with a gun. Finally, in 2021, Rahimi shot a gun in the air when his friend’s credit card was declined (Note 2). Restraining orders are unlikely to work against the people who are truly dangerous. Someone who is willing to commit murder is already facing either a life sentence or the death penalty. The additional penalties for illegally obtaining a gun or violating a protective order are unlikely to provide marginal deterrence. Despite a protective order, Mr. Rahimi still illegally obtained and used a firearm. In any application of a restraining order, there are false positives and false negatives. Restraining orders mistakenly applied to law-abiding individuals will likely be effective in disarming them, so that they will not be able to defend themselves and others. As just noted, restraining orders might not impose a real marginal penalty on those who are truly dangerous. Thus, the law is most likely to restrain only those who are most law-abiding and fail to restrain the most dangerous. What are the merits of this law that will be lost in the protection of Second Amendment rights if the Supreme Court finds it unconstitutional? In this paper, we attempt to assess the benefits of banning individuals who are subject to domestic violence protection orders from possessing firearms. That is, we are assessing the costs of declaring 18 U.S.C. § 922(g)(8) unconstitutional. www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 15 Published by SCHOLINK INC. 2. Method 2.1 Econometric Methodology We measure the strength of state Domestic Violence Protective Orders (DVPO) by the duration of the order. The 922 law enhances the state domestic violence protection order. It has no effect in the absence of a state protection order. The 922 law passed in 1994. We create a dummy variable to measure its effect, taking the value zero for years before 1994 and the unit value for the remaining years. We then create two test variables by multiplying the temporary order duration and the final order duration variables by the dummy variable for 1994 and later. We are, therefore, estimating the additional effect of the 922 law given an existing state protection order. This methodology allows us to estimate a Two-Way Fixed-Effects (TWFE) panel data model where we include the year dummies to control for things like recessions, the passage of other federal laws, and anything else that affects all states in a given year. It also makes the necessary assumption of independence across states more likely. A DVPO is intended to protect the applicant against potential violence committed by a domestic partner. While a person of any gender can request such an order, women are the primary users. As such, we investigate several different outcomes, domestic homicides, domestic homicides of women, domestic homicides committed by firearms, and femicides by a domestic partner using firearms. Since these are count variables, we use the TWFE negative binomial model (Cameron & Trivedi, 1998, pp. 70-79). The negative binomial is a generalization of the Poisson distribution in that it does not require the mean and variance of the distribution of the dependent variable to be equal. The typical case is that the variance is greater than the mean, called overdispersion. We tested for overdispersion in each of the models reported below. There was significant overdispersion in every case, justifying the negative binomial model. The model is as follows. 1 2 3 4 log( ) ( ) ( ) it i t it it it it it it it it y tdvpo fdvpo D tdvpo D fdvpo X u              (1) The αi are the state fixed effects, the δt are the year fixed effects, Dit is the dummy variable indicating that the year is 1994 or later, tdpvoit is the duration of the state’s temporary protection order in year t, fdvpoit is the duration of the state’s final DVPO in that year, Xit is a matrix containing observations on the control variables. The primary coefficients of interest are β3 and β4 indicating the enhancement effect of 922. However, the coefficients on the temporary and final protection order durations (β1 and β2) may also be of interest. The choice of control variables to include in the model is crucial. If we leave out a potentially relevant variable, it results in omitted variable bias. Including an irrelevant variable whose true coefficient is zero, does not bias the estimates, but the estimates are inefficient. The inefficiency could be so bad that a relevant variable could have a standard error so large that it inaccurately appears to be insignificant. www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 16 Published by SCHOLINK INC. According to Rao (1971) and Wallace (1964), if the true t-ratio for a given variable is less than one in absolute value, dropping this potentially irrelevant variable from the original (general) model will reduce the mean squared error (variance plus squared bias) of all the remaining coefficients in the resulting (specific) model. Accordingly, we start with a general model including many potentially relevant variables. We then drop all control variables with absolute t-ratios less than one and test them as a group for significance using a standard F-test. If the F-test is not significant, we are justified in dropping the insignificant variables. If the F-test is significant, we go back and find the variable or variables making it significant and add them back into the model. We continue until the F-test is insignificant. If the resulting specific model has any t-statistics less than one in absolute value, we drop them, add them to the F-test, and test for significance. If the F-test is insignificant, we re-estimate the new, reduced, specific model. We repeat until the specific model has no t-ratios less than one in absolute value. The resulting model is the specific model and, while it may have some bias, the bias will be small. The specific model will be more efficient, with presumably more precise estimates, and have lower mean squared errors, than the general model. This general-to-specific methodology (Hendry, 1995) has been used successfully in many applications, e.g., Owen and Weatherston (2004), Muelbauer and Nunziata (2004), Rao and Singh (2006), and Reade (2007). We rely on this modeling strategy to estimate the true data generating process as closely as possible. It is unlikely that the federal government passed the law banning individuals subject to a DVPO from possessing firearms because of a contemporaneous increase in deaths of individuals otherwise protected by state protection orders. We assume that the policy variables are exogenously determined. That means that we can avoid the use of instrumental variables. However, since some of the control variables could be simultaneously determined with the dependent variable, we lag the potentially endogenous control variables. 2.2 Data Observations on the dependent variables are taken from Kaplan (2021). Jacob Kaplan has produced a convenient database for the FBI’s Supplementary Homicide Reports (SHR). The SHR data for 1980-1983 are missing and most of the data for the control variables end in 2018. Consequently, our primary results are derived from an unbalanced panel consisting of 50 states. In the robustness section we restrict the dataset to 1984-2018 to form a balanced panel. We defined a domestic homicide as one in which the first victim was related to the first offender as a husband, wife, boyfriend, girlfriend, common-law husband, common-law wife, ex-husband, ex-wife, or in a homosexual relationship. We then determined if the victim was female and if the offender used a firearm. For the policy variables, we examined the law codes of all fifty states from 1970 to 2023. There are two types of protective orders. Temporary or emergency orders are usually issued without the participation of the respondent. These are usually short, averaging 27 days (median 14 days), but they can be as long as two years. The final order, which replaces any temporary order, is longer, averaging four years with www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 17 Published by SCHOLINK INC. a median of one year. However, many states issue permanent orders, which could be modified or rescinded. To translate these into numerical values, we had to make several assumptions. For example, since we don’t know the respondent’s age or the probability of the order being appealed or otherwise reduced or rescinded, we assume that a permanent order lasts 25 years. The online appendix lists all the assumptions. Since the numerical values are necessarily arbitrary, we did robustness checks with these long-duration order values doubled and halved to see if the results changed. The results are not sensitive to these assumptions. The control variables are listed in Table 1 along with the means of the dependent variables and the policy variables of interest. Table 1. Means and Other Statistics Variable N Mean St. Dev Min Max Domestic murders 1587 33.73 43.53 0 344 Domestic femicides 1587 23.88 30.18 0 217 Dom gun murders 1587 19.59 27.24 0 241 Dom gun femicides 1587 14.37 19.02 0 141 Final order, years 1587 3.98 7.40 0.05 42.48 Temp order, days 1587 27.13 73.38 0.28 730 Final*922 dummy 1587 3.55 7.26 0 42.48 Temp*922 dummy 1587 21.19 68.85 0 730 Prison pc 1587 333.90 154.32 32.84 871.62 Police pc 1587 277.60 50.41 161.67 522.74 Abortion 1587 1.51 1.31 0 6.02 Unemployment rate 1587 5.65 1.88 2.30 14.80 Employment pc 1587 57.90 5.33 37.99 73.07 Military pc 1587 1.01 0.90 0.27 6.65 Construction pc 1587 3.33 0.75 1.36 6.90 Alcohol pc 1587 1.91 0.45 0.89 4.04 Crack 1587 1.41 1.10 -0.70 7.78 Density 1587 4.31 13.54 0.08 111.34 Income pc 1587 16.80 3.48 9.16 30.45 Poverty rate 1587 12.89 3.71 2.90 27.20 Welfare pc 1587 249.58 89.65 0.33 548.53 Gun suicide/suicide 1587 61.10 12.73 0 100 Executions pc 1587 0.87 3.08 0 40 Percent population 15-34 1587 28.92 2.72 23.37 39.81 Percent black men 15-34 1587 1.66 1.44 0.07 5.90 www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 18 Published by SCHOLINK INC. Police, prison, and the death penalty all have deterrent roles. Prison and executions also have incapacitation effects. These variables are all in per capita rates. We also lag them to avoid simultaneity. The effective abortion rate for murder from Donohue and Levitt (2020) is included because there is some evidence that legalized abortion could affect crime rates. Unfortunately, the Donohue and Levitt (2020) sample ends in 2014 and computation of the effective abortion rate is complicated. However, it is easy to extrapolate the series since it is a smooth trend, and we only need to extrapolate four years. We include the proportion of the population between 15 and 34 and the proportion of black males 15-34 because relatively young men largely commit murder. We have three employment variables, total employment per capita, indicating the amount of legitimate employment available; military employment, since the military concentrates many young men in certain areas but also sends many young men out of the country; and construction employment per capita because the construction industry also concentrates large numbers of young men. The denser the population, the more interactions among the inhabitants, some of which could cause domestic murder. For this reason, we include population density as a control variable. We include per capita consumption of alcohol because alcohol can reduce inhibitions. We include the unemployment rate, real personal income per capita, the poverty rate, and real welfare payments per capita, any of which could affect domestic murder by altering the stresses associated with domestic life. A possibly important factor in any crime equation using historical data is the emergence of crack cocaine in the 1980’s. The resulting large increase in the supply of cocaine caused turf battles among drug gangs, increasing murder rates. We control for the crack cocaine epidemic by including the Fryer et al. (2013) crack index, a combination of indicators of cocaine use compiled by Fryer and his colleagues for the period of the crack epidemic. There are continuous values for 1981 to 2000 for each state. We set pre-1981 values at the 1981 levels and post-2000 values at the 2000 levels. Finally, we include population as an exposure variable. 3. Results The next two tables present the primary results. Table 2 shows the effects of the federal 922 law on domestic murder and domestic femicide. The temporary and final domestic violence protection orders insignificantly change domestic murders or domestic femicides. Table 3 shows the law’s effects on domestic gun murder and domestic gun femicide. Here we again find that neither temporary nor final domestic violence protection orders significantly reduce domestic gun murders or domestic gun femicides. The effect of the 922 law working through temporary orders is a small, but significant, increase in domestic gun murders. The estimates for domestic gun femicides show a similarly small effect but are not consistently significant. The effect of the 922 law working through final orders shows a very small and insignificant decline in both domestic gun murders and domestic gun femicides. Overall, 18 U.S.C. § 922(g)(8) has not significantly reduced domestic murders, domestic www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 19 Published by SCHOLINK INC. femicides, domestic gun murders, or domestic gun femicides. Table 2. Results for Domestic Murder and Domestic Femicide Domestic Murder Domestic Femicide Variables General Specific General Specific Temporary order, days 1.000 1.001 1.001 1.001 (0.602) (0.862) (1.031) (1.364) Final order, years 0.993 0.991 0.993 0.991 (-1.183) (-1.250) (-1.101) (-1.165) 922 dummy * temp days 1.000 1.000 1.000 1.000 (0.556) (1.350) (-0.432) (0.121) 922 dummy * final years 0.997 0.994 0.998 0.994 (-0.651) (-1.103) (-0.423) (-1.115) Prisoners per capita lagged 1.000 1.000 1.000 (-1.356) (-1.338) (-0.653) Police per capita lagged 1.000 1.000 (-0.302) (0.0198) Abortion 0.984 0.958 (-0.282) (-0.788) Unemployment 1.010 1.014 (0.772) (1.125) Employment 0.993 0.995 (-0.460) (-0.300) Military 1.007 1.039 (0.0536) (0.305) Construction 1.103 1.078 1.087 1.041 (1.798) (1.760) (1.506) (1.044) Alcohol 1.503* 1.446** 1.335 1.386* (2.172) (2.810) (1.541) (2.551) Crack 1.042 1.076 1.042 1.079 (1.411) (1.673) (1.376) (1.666) Density 0.995 0.985 0.982* (-0.485) (-1.540) (-2.135) Income 1.004 1.007 (0.192) (0.345) Poverty 1.011 1.014 1.007 www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 20 Published by SCHOLINK INC. (1.221) (1.665) (0.779) Welfare 1.000 1.000 (-0.666) (-0.0126) Gun suicide/suicide lagged 0.997 0.997 0.997 0.998 (-1.671) (-1.450) (-1.406) (-1.072) Executions pc lagged 1.000 1.001 (-0.0873) (0.414) Pct pop 15-34 0.982 0.976 (-0.707) (-0.941) Pct black men 15-34 1.751*** 1.613*** 1.558*** 1.471** (3.813) (3.378) (3.357) (3.014) Note. ***p<.001, p<**.01, p<*.05; negative binomial model; coefficients are incidence rate ratios; t-ratios in parentheses; standard errors clustered on states; state and year effects estimates are suppressed; complete results, programs and data are available at . Table 3. Domestic Gun Murder, Domestic Gun Femicide Domestic gun murder Domestic gun femicide Variables General Specific General Specific Temporary order, days 1.000 1.001 1.000 1.000 (0.274) (0.703) (0.323) (0.486) Final order, years 0.994 0.992 0.991 0.991 (-0.965) (-1.104) (-1.402) (-1.351) 922 dummy * temp days 1.001* 1.001** 1.001* 1.001 (2.083) (2.588) (2.258) (1.839) 922 dummy * final order years 0.993 0.991 0.996 0.996 (-1.329) (-1.654) (-0.786) (-0.826) Prisoners per capita lagged 1.000 1.000 (-0.316) (0.400) Police per capita lagged 0.999 0.999 (-0.737) (-0.434) Abortion 0.962 0.956 (-0.614) (-0.826) Unemployment 1.013 1.006 (0.808) (0.411) Employment 0.990 0.997 (-0.461) (-0.157) www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 21 Published by SCHOLINK INC. Military 0.974 0.978 (-0.179) (-0.171) Construction 1.106 1.088 (1.562) (1.360) Alcohol 1.648* 1.664*** 1.470 1.589*** (2.317) (3.373) (1.850) (3.524) Crack 1.072* 1.109* 1.084* 1.091** (2.041) (2.242) (2.314) (2.768) Density 0.997 0.991 0.984 (-0.270) (-1.075) (-1.537) Income 0.994 0.987 (-0.294) (-0.698) Poverty 1.015 1.016* 1.010 (1.602) (1.995) (1.143) Welfare 1.000 1.000 (-0.243) (0.376) Gun suicide/suicide lagged 0.997 0.998 (-1.112) (-0.995) Executions pc lagged 0.996 0.995 0.998 (-1.356) (-1.889) (-0.727) Pct pop 15-34 0.979 0.968 0.972 (-0.660) (-1.045) (-1.032) Pct black men 15-34 1.697** 1.525* 1.542** 1.571** (3.192) (2.458) (2.843) (2.997) Note. ***p<.001, p<**.01, p<*.05; negative binomial model; coefficients are incidence rate ratios; t-ratios in parentheses; standard errors clustered on states; state and year effects estimates are suppressed; complete results, programs and data are available at . 4. Robustness Checks An alternative method for estimating the impact of a crime policy is to investigate the policy’s effect on the per capita domestic murder rate. For this analysis, we apply the two-way fixed-effects model to the rate of domestic murders per one million population. Table 4 presents the results. The estimates are very similar to those shown in Tables 2 and 3. The federal 922 law does not significantly affect domestic murder or domestic femicide. The small positive effect for domestic gun murders which was significant in Table 3 is not significant in the per capita fixed-effects model. As reported in Table 3, there is a significant small positive effect on domestic gun femicide for temporary protective orders, but only in the more efficient specific model. www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 22 Published by SCHOLINK INC. Table 4. Estimates Using Per Capita Data Domestic murder Domestic femicide General Specific General Specific Temporary order, days 0.034 0.038 0.011 0.0071 (0.87) (1.03) (0.40) (0.25) Final order, years -0.36 -0.37 -0.18 -0.19 (-1.08) (-1.19) (-0.85) (-0.82) 922 dummy * temp days -0.015 -0.016 -0.0061 0.0046 (-1.13) (-1.33) (-0.59) (0.50) 922 dummy * final order years 0.033 0.028 0.049 -0.075 (0.14) (0.12) (0.31) (-0.44) Domestic gun murder Domestic gun femicide General Specific General Specific Temporary order, days 0.026 0.032 0.0044 -0.0025 (1.05) (1.31) (0.24) (-0.14) Final order, years -0.21 -0.27 -0.24 -0.22 (-0.79) (-1.01) (-1.41) (-1.16) 922 dummy * temp days 0.00031 -0.0012 0.011 0.027** (0.032) (-0.13) (1.39) (3.46) 922 dummy * final order years -0.16 -0.12 0.0034 -0.070 (-0.78) (-0.62) (0.027) (-0.49) Note. ***p<.001, p<**.01, p<*.05; two-way fixed-effects model; control variables and state and year effects estimates are suppressed; standard errors clustered on states; full results, programs, and data available at . As noted above, there is some uncertainty as to how to translate the duration of indefinite or permanent final orders into numbers that can be entered into a dataset. We investigate the sensitivity of the results to these arbitrary values by doubling and halving the values of any orders over 10 years and re-running the negative binomial regressions. The results are reported in Table 5. www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 23 Published by SCHOLINK INC. Table 5. Doubling and Halving Long Sentences 2X Half X Domestic murder Domestic murder General Specific General Specific Temporary order, days 1.00 1.00 1.00 1.00 (0.57) (0.84) (0.65) (0.89) Final order, years 1.00 1.00 0.98* 0.99 (-0.88) (-1.07) (-2.05) (-1.70) 922 dummy * temp days 1.00 1.00 1.00 1.00 (0.61) (1.42) (0.44) (1.24) 922 dummy * final order years 1.00 0.99 1.00 0.99 (-0.87) (-1.27) (-0.40) (-0.89) Domestic femicide Domestic femicide General Specific General Specific Temporary order, days 1.00 1.00 1.00 1.00 (1.00) (1.34) (1.10) (1.42) Final order, years 1.00 1.00 0.98 0.99 (-0.82) (-1.02) (-1.86) (-1.55) 922 dummy * temp days 1.00 1.00 1.00 1.00 (-0.39) (0.19) (-0.55) (0.0044) 922 dummy * final order years 1.00 0.99 1.00 0.99 (-0.65) (-1.27) (-0.22) (-0.92) Domestic gun murder Domestic gun murder General Specific General Specific Temporary order, days 1.00 1.00 1.00 1.00 (0.23) (0.67) (0.35) (0.76) Final order, years 1.00 1.00 0.98 0.98 (-0.55) (-0.87) (-1.82) (-1.77) 922 dummy * temp days 1.00* 1.00** 1.00 1.00* (2.14) (2.66) (1.95) (2.46) 922 dummy*final order years 0.99 0.99 1.00 0.99 (-1.52) (-1.86) (-0.72) (-1.14) Domestic gun femicide Domestic gun femicide General Specific General Specific Temporary order, days 1.00 1.00 1.00 1.00 (0.28) (0.44) (0.38) (0.55) www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 24 Published by SCHOLINK INC. Final order, years 1.00 1.00 0.98* 0.98* (-1.04) (-1.01) (-2.23) (-2.24) 922 dummy * temp days 1.00* 1.00 1.00* 1.00 (2.34) (1.91) (2.08) (1.71) 922 dummy*final order years 0.99 0.99 1.00 1.00 (-1.02) (-1.04) (-0.46) (-0.52) Note. ***p<.001, p<**.01, p<*.05; indefinite long sentences for final orders are doubled in the first two columns and halved in columns three and four; negative binomial model; coefficients are incidence rate ratios; t-ratios in parentheses; standard errors clustered on states; control variables, state and year effects estimates are suppressed; complete results, programs and data are available at . The results are virtually the same as those reported in Tables 2 and 3 and they are almost identical for doubling and halving the order length. The results appear to be robust to assumptions concerning final order duration. The SHR data has missing values for all states from 1980-1983. The resulting panel is unbalanced. To get a balanced panel, we estimated the negative binomial model using data from 1984-2018. The results are reported in Table 6. Table 6. Using Balanced Panel, 1984-2018 Domestic murder Domestic femicide General Specific General Specific Temporary order, days 1.000 1.000 1.001 1.001 (0.457) (0.445) (0.875) (0.944) Final order, years 0.992 0.992 0.992 0.992 (-1.255) (-1.192) (-1.224) (-1.094) 922 dummy * temp days 1.000 1.000 1.000 1.000 (0.530) (1.306) (-0.454) (-0.0547) 922 dummy * final order years 0.997 0.996 0.999 0.997 (-0.660) (-0.819) (-0.345) (-0.749) Domestic gun murder Domestic gun femicide General Specific General Specific Temporary order, days 1.000 1.000 1.000 1.000 (0.140) (0.224) (0.222) (0.400) Final order, years 0.994 0.993 0.990 0.991 (-1.040) (-1.016) (-1.520) (-1.447) www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 25 Published by SCHOLINK INC. 922 dummy * temp days 1.001 1.001* 1.001* 1.001 (1.948) (2.453) (2.145) (1.702) 922 dummy * final order years 0.993 0.993 0.997 0.996 (-1.357) (-1.451) (-0.707) (-0.838) Note. ***p<.001, p<**.01, p<*.05; sample limited to 1984-2018; negative binomial model; coefficients are incidence rate ratios; t-ratios in parentheses; standard errors clustered on states; state and year effects estimates are suppressed; complete results, programs and data are available at . The results are almost identical to those reported in Tables 2 and 3. 5. Discussion We have estimated the enhancement effect of federal law 18 U.S.C. § 922(g)(8) using a panel of fifty states over 38 years. We have subjected the analysis to a variety of robustness checks. The results are remarkably robust. We find that 18 U.S.C. § 922(g)(8) has no significant effect on domestic murder or domestic femicide. We find some evidence that the law positively affects domestic gun murder and domestic gun femicide. This could be caused by a Peltzman (1975) effect, where individuals who feel safe engage in risky behavior. Overall, we conclude that 18 U.S.C. § 922(g)(8) does not significantly reduce domestic murder, domestic femicide, domestic gun murder, or domestic gun femicide. Complete results, programs, and data are available in the online appendix . References Cameron, A. C., & Trivedi, P. K. (1998). Regression Analysis of Count Data. Cambridge, UK. https://doi.org/10.1017/CBO9780511814365 Donohue, J. J., & Levitt, S. D. (2020). The Impact of Legalized Abortion on Crime over the Last Two Decades. American Law and Economics Review, 22, 241-302. https://doi.org/10.1093/aler/ahaa008 Fryer, R. G., Heaton, P. S., Levitt, S. D., & Murphy, K. M. (2013). Measuring Crack Cocaine and Its Impact. Economic Inquiry, 51, 1651-1681. https://doi.org/10.1111/j.1465-7295.2012.00506.x Hendry, D. F. (1995). Dynamic Econometrics. Oxford University Press, Oxford, UK. https://doi.org/10.1093/0198283164.001.0001 Kaplan, J. (2021). Jacob Kaplan’s Concatenated Files, Uniform Crime Reporting (UCR) Program Data, Supplementary Homicide Reports (SHR), 1976-2020. Ann Arbor, MI, Inter-university Consortium for Political and Social Research [distributor], 2021-09-22. https://doi.org/10.21428/cb6ab371.f1bc7339 https://doi.org/10.1017/CBO9780511814365 https://doi.org/10.1093/aler/ahaa008 https://doi.org/10.1111/j.1465-7295.2012.00506.x https://doi.org/10.1093/0198283164.001.0001 https://doi.org/10.21428/cb6ab371.f1bc7339 www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 6, No. 3, 2023 26 Published by SCHOLINK INC. Muelbauer, J., & Nunziata, L. (2004). Forecasting (and explaining) US Business Cycles, CEPR Discussion Papers 4584. Owen, D., & Weatherston, C. R. (2004). Uncertainty of outcome and super 12 attendance, application of a general-to-specific modeling strategy. Journal of Sports Economics, 5, 347-370. https://doi.org/10.1177/1527002503259062 Peltzman, S. (1975). The effects of automobile safety regulation. Journal of Political Economy, 83, 677-726. https://doi.org/10.1086/260352 Rao, B. B., & Singh, R. (2006). Demand for money for Fiji with PcGets. Applied Economics Letters, 13, 987-991. https://doi.org/10.1080/13504850500425402 Rao, P. (1971). Some notes on misspecification in multiple regression models. The American Statistician, 25, 37-39. https://doi.org/10.1080/00031305.1971.10477302 Reade, J. J. (2007). Modeling and forecasting football attendances. Oxonomics, 2, 27-32. https://doi.org/10.1111/j.1752-5209.2007.00015.x Wallace, T. D. (1964). Efficiencies for stepwise regressions. Journal of the American Statistical Association, 59, 1179-1182. https://doi.org/10.1080/01621459.1964.10480758 Notes Note 1. UNITED STATES OF AMERICA, PETITIONER v. ZACKEY RAHIMI (https,//www.supremecourt.gov/DocketPDF/22/22-915/259334/20230317174308399_Rahimi%20Pet% 20-%20final.pdf, p. 66a.). Note 2. John Fitze, “How a Second Amendment case at the Supreme Court is putting gun rights groups in a jam”, USA Today, July 12, 2023 (https,//www.usatoday.com/story/news/politics/2023/07/12/guns-supreme-court-second-amendment-ra himi/70383454007/). https://doi.org/10.1177/1527002503259062 https://doi.org/10.1086/260352 https://doi.org/10.1080/13504850500425402 https://doi.org/10.1080/00031305.1971.10477302 https://doi.org/10.1111/j.1752-5209.2007.00015.x https://doi.org/10.1080/01621459.1964.10480758