by Rodrigo Schneider
[7 minute read]
In 2003, Brazilian Congress approved legislation that banned the carrying of concealed weapons nationwide and provided for a voter referendum 22 months later regarding whether to ban the sale of all firearms in Brazil. Whether this restrictive law was effective in reducing crimes is still controversial and I use a statistical design to identify the causal effects of this legislation and provide more explicit evidence to elucidate this law’s consequences.
In the early 2000s, more than 30,000 gun-related homicides occurred in Brazil every year. This number was much smaller in the 1980s but sharply increased in the 1990s. Motivated by this dramatic increase in the number of firearm-related deaths in Brazil, legislators passed federal firearm regulations in December 22nd, 2003 (Law number 10.826), in the form of the Estatuto do Desarmamento (Disarmament Statute). The legislation prohibited citizens from carrying a gun outside of their residences or places of business; it provided exemptions for hunters (sporting or subsistence), private security employees, and police officers. The penalty for illegal possession (or carrying) increased from an incarceration period of one to three months to two to four years. Finally, the statute made obtaining a gun permit more expensive and imposed more stringent requirements that made the process more restrictive. This package of measures was enacted to decrease gun violence.
An essential and unique feature of the legislation was its 35th section, which set the stage for a national referendum to take place in October 2005 (22 months after the initial legislation was passed into law), to allow Brazilian citizens to vote on an even more restrictive weapons laws. The law put forward in the referendum stipulated that the sale of any guns and ammunition would be completely prohibited in the country (again, with exceptions for hunters and those with security-related jobs). More specifically, voters were asked the following question: Should the commerce of firearms and ammunition be prohibited in Brazil? Therefore, the referendum did not propose to change the previously passed legislative statute, which prohibited the carrying of concealed weapons, but it proposed to go further, by prohibiting the sale of all firearms.
By using a statistical design, commonly adopted to identify the impact of a policy in a reliable way (called regression discontinuity design), I found that in 2004, one year after the law was implemented, gun-related homicides decreased by 10.8 percent, with the reduction especially pronounced among young black males living in high-crime areas. Other crimes involving guns also declined (e.g. armed robbery decreased by 7.7%), while crimes that did not involve guns were unaffected. Enrollment in adult education courses disproportionately increased in areas that saw the biggest drop in gun-related crimes. 
As mentioned above, 22 months after the prohibition of concealed carry, Brazilian voters decided on a referendum whether owning a gun, even inside your residence or place of business, should be prohibited as well. Statistical analysis of this referendum, which was defeated by a wide margin (64% voted against the prohibition), shows higher voter turnout and stronger support for the complete weapons ban in the areas that had experienced the most significant decline in gun-related homicides.
Analyzing the concealed carry prohibition, I find that gun-related homicides decreased by 10.8% in the year following the enactment of the law (i.e., 2004), which corresponds to 3,900 lives being spared in one year. As figure 1 shows, the law reduced gun-related homicides but did not affect non-gun-related homicides mitigating concerns about the existence of other factors related to the crime that could be simultaneously changing with the law and potentially driving my results. In Brazil, estimations of the value of statistical life vary from $0.77 million to $6.1 million. Therefore, using the most conservative value of statistical life allows me to make the following claim: The prohibition of the right to carry concealed weapons generated an economic value of $3 billion in one year. This estimation does not consider other gains generated by the law such as a reduction of 12% in gunshot wounds, 8% in robberies and an increase in enrollment in adult education courses concentrated in areas where the concealed carry prohibition was most effective.
Figure 1 – Effect of the concealed carry prohibition on total homicides, gun-related homicides and non-gun-related homicides per 100,000 people
Notes: Figure 1 shows three time-varying functions using a 48 months’ bandwidth and a vertical red line representing the cutoff point (January 2004). The solid line is fitted separately on each side of the threshold, and the dashed line represents the 95% confidence interval. The scatter plots show monthly averages. I regress the predicted residuals after regressing my dependent variables on calendar months, monthly rainfall and temperatures to take seasonality into account.
Although the concealed carry prohibition was highly effective and generated large welfare gains, the referendum on gun prohibition failed to pass. Only 36% of voters supported the gun prohibition, which required a simple majority to pass. Areas that experienced more significant decreases in gun-related homicides showed higher levels of voter turnout and support for the referendum that proposed a complete firearm ban. Figure 2 illustrates this argument showing the neighborhoods of the municipality of São Paulo: high-crime neighborhoods showed larger support for the gun ban. This result provides a suggestive hint as to why gun control legislation may be difficult to pass: Welfare gains may be concentrated in a small and less privileged share of the population that does not have enough votes, organizational strength or even influence in politics.
Figure 2 – Relationship between voting for the prohibition and vulnerability index
Notes: The dashed line represents the least square estimation of the relationship between the residuals of the linear regression of support for gun control on population and income and the residuals of the linear regression of the vulnerability index on population and income. The regression considers all 47 neighborhoods of the São Paulo municipality for which the Supreme Electoral Court (TSE) provides information on.
After 14 years of the enactment of the concealed carry ban, Brazil remains as the country with the largest number of gun-related homicides in the world. This can give us suggestions that law enforcement is a complement to any legislation and that, the law by itself, cannot solve our gun violence issues. Examples like the gun buyback program in Australia that reduced gun-related deaths by about 80% (8 times larger than the effect measured in the Brazilian gun carry ban) can show us that gun control policies, when enacted in countries with better police enforcement and easier border control than ours, can be even more effective. An additional lesson that the literature related to gun control can teach us, especially from empirical examinations in the United States, is that allowing people to carry a gun only increase crimes. Therefore, laws that aim to allow Brazilians citizens to carry a gun, as the proposed bill PL 3722/2012, should not be expected to solve our gun violence problem, but only make it worse.
The investigation of the Brazilian concealed carry prohibition can also inspire the analysis of other laws attempting to reduce crimes. As mentioned above, the concealed carry ban disproportionately affected males as 94% of gun-related homicides’ victims are men. Other Brazilian laws, such as the Maria da Penha law, were focused in reducing crime against women and research showing whether it was effective and which part of society gained the most from it are not only welcome but necessary. Especially, because, as the referendum outcome suggested, Brazilian laws although effective, may have its gains concentrated in the most vulnerable part of the society. While this seems good from the point of view of an egalitarian society, as the most vulnerable people are likely to have their voices ignored, these legislations might not be appreciated and disregarded by the majority compromising their long-run effects.
 For details about my work methodology and findings, see: https://drive.google.com/file/d/0B1QCQasYG8iHaWJVZ0Z3VDJ2eEE/view
 For instance, in 2003, there were 36,115 gun-related homicides (calculated using data from Brazilian Health Ministry).
 This penalty is harsher than most of the ones applied in the United States, where most states punish possession of gun without permit as a misdemeanor. For instance, in New York, possession without permit is punishable by up to one year in prison, a fine of up to $1,000, or both
 An applicant should have no criminal record, be employed, show proof of residence, pay a fee close to $1,000 attend a gun safety course, and pass a psychological exam.
 To identify the impact of the gun carry ban on adult education enrollment, I used a statistical method called difference-in-differences. It compares municipalities where the effects of the concealed carry ban were larger to municipalities where they were smaller, before and after the law. The results shows that places where the law had its larger effects also experienced a sharper increase in adult education enrollment. This was driven by male enrollment. I examined female enrollment as a placebo test and found no statistical effect, which should be expected as the law mostly affected males (94% of gun-related homicides victims were males). The most plausible mechanism explaining this result is that the law, by increasing the cost of being a criminal, decreased the opportunity cost of education.
 The economic value of the regulation can be estimated using the literature on the value of a statistical life.
 Ortiz, R., Markandya, A., and Hunt, A. (2009). Willingness to Pay for Mortality Risk Reduction Associated with Air Pollution in São Paulo. Revista Brasileira de Economia, 63, 3-22.
 Retrieved from my paper (https://drive.google.com/file/d/0B1QCQasYG8iHaWJVZ0Z3VDJ2eEE/view)
 Retrieved from my paper (https://drive.google.com/file/d/0B1QCQasYG8iHaWJVZ0Z3VDJ2eEE/view)
Rodrigo Schneider is Brazilian, originally from Itapetininga city in São Paulo state. He is a Ph.D. candidate in economics at the University of Illinois at Urbana-Champaign. His research interests lie at the intersection of public policy, development and political economy.