Introduction

According to the National Institute of Justice, 467,321 people fell victom to a crime involving a firearm in 2011. 68% of all murders, 41% of robbery offenses, and 21% of aggravated assaults nationwide involved the use of a gun (National Institue of Justice, 2019). In 2018, an estimated 368.9 violent crimes were committed per 100,000 inahabitants in the United States (FBI National Press Office, 2019). However, the overall trend of violent crime in the last ten years has been of decreasing slope.

Aproximately four in ten U.S adults say they live in a gun-owning household. Among these households, 30 percent state that they own their own gun (Gramlich & Schaeffer, 2019). As of 2019, all 50 states and Washington D.C. allow citizens to carry a concealed firearm (“State-By-State Concealed Carry Permit Laws”, 2019). However, the law on which a citizen is allowed to openly carry their firearm varies among states.

Gun control has been a pressing issue in American society for years. Political parties have been producing polarized legislastion that have been controversial to the public eye. Parties have seperate views but ultimately share the same end goal; to reduce the amount of harm to society and maximize the safety of the American people.

Carrying a concealed weapon can quickly resolve serious threats and de-escalate tragic events. Citizens are able to take safety into their own hands with their own actions. Problems can arise from this as well. Just as quickly, a non-threating situation can be escalted and lead to unjust deaths. Weapons falling into the wrong hands can lead to the taking of innocent lives.

This paper quantifies the effectiveness of the different open carry gun control laws, basing the measurement of effectiveness off of violent crimes commited. Nationally represented surveys and census information is utilized via regression to create findings on whether the laws can be considered effective in reducing violent crimes.

Background and Rationale

Every state in the United States allows citizens to conceal a firearm. Laws differentiate when it comes to allowing citizens to openly carry that firearm. To open carry is defined as “the act of carrying a gun in public where it is not hidden from common observation” (“Open Carry”, n.d.). States have different federal laws regarding the definition of “common observation” and differ in what is required to be able to openly carry a fire arm. Some states completely ban the practice of open carry.

Gun Level Number of States Percentage
1 27 56%
2 13 12%
3 4 20%
4 6 12%

Six states prohibit a citzen from openly carrying a hand gun. Twenty-seven states allow citizens to openly carry a hand gun without the need of a permit or liscence. The remaining seventeen states allow open carry as long as the citizens obtains a permit. However, of these seventeen, four states have local authorities regulate. This means that local governments may pass gun laws more restrictive than the state. This information was gathered from the United States Concealed Carry Association.

For example, the state of Illinois follows the law of banning open carry completely. The law prohibits any person from knowingly carrying a firearm on public place or land. Citizens of Illinois are allowed to conceal a weapon but can only carry it on their own personal land, in the home of a friend’s with their permission, or on certain hunting grounds (USCCA, 2019). Other states that ban the practice of open carry generally follow these same rules.

For a state such as Alabama which allows open carry without a permit, anyone who is at least 18 years old and legally entitled to carry a firearm is allowed to openly carry a weapon. States such as Minnesota which allows a permit, the citizen must register and obtain a permit in order to openly carry there weapon. States such as Oregon allow open carry with a permit, but also regulate and restrict certain actions. The sheriff is allowed discretion and is allow to deem applicants reasonably likely to be a danger to themselves or others.

Previous Literature

Previous research in the United States has looked into the different levels of gun control legislation and found that certain levels of gun control laws prove to be significant while other levels do not (Kleck, Kovandzic, & Bellow, 2016). Other research ranked legislation on an index from 1-100 based on the varying strictness. Regression analysis showed that gun control could not be considered a significant predictor of violent crime, therefore concluded that there was not enough evidence that gun control reduced or increased crime rates. (Moorhouse & Wanner, 2006).

Study Objectives

Previous experiments looked into 19 major gun control laws and ranked gun control laws on an index. This paper zones in on open carry laws specifically and will place the laws on an index from 1-4, representing the 4 different open carry laws. This will help draw conclusions to the open carry debate rather than the more broad debate of overall gun control. Aside from the differences of laws analyzed, this paper will dive into data statewide, rather than citywide. Observations will be of states throughout time rather than focusing on cities with over 25,000 population.

Empirical Strategy

This study was designed to quantify the effect of open carry gun laws measuring on a basis of violent crimes committed. States were given a number from 1 through 4 based on how strict their open carry legislation is. The four levels consisted of states that allowed open carry without requiring a permit (level 1), states that allowed open carry with a permit (level 2), states that allowed open carry with a permit and heavily regulated laws (level 3), and states that did not allow open carry whatsoever (level 4). These four levels were then ran in a regression to see the effect of varying gun laws on violent crimes commited.

Rates were utilized to make numbers more legible and interprable. It was decided that violent crimes would be best measured on a per 1000 rate. This helps control for population size as well. As population increases, crimes should as well, so dividing the total crimes by the population will help control for the effect of population. Therefore, the beginning regression consisted of predicting violent crime rates per 1000 by the different levels of gun control. This is represented by the following equation:

\(Violent CrimesperThousand_{i} = \beta_{0} + \beta_{1}GunLaws_{i} + \epsilon_{i}\)

It was deemed that this equation was victim of omitted variable bias. For example, a higher poverty rate may lead to an increase in violent crimes as these people are willing to risk deviant behavior as they have less to lose. A low GDP producing area may see less development and have less money to spend on resources to prevent crime, therefor may see an increase in crime. As weather increase, more civilians will spend time out and about, leading to more interactions, which may in term lead to more conflict and crime. Each of these coefficeints are expected to be positive. Therefore population, poverty rates, GDP, and climate were added to the equation, as these must be held constant throughout due to the fact that these also influence the amount of violent crimes committed in an area. Gun laws, population, GDP, climate, poverty, and violent crimes were all measured individually in each state from the years 1997-2014. The new population regression equation stands at:

\(Violent CrimesperThousand_{i} = \beta_{0} + \beta_{1}GunLaws_{i} + \beta_{1}PovertyRate_{i} + \beta_{1}Climate_{i} + \beta_{1}GDPperCap_{i} + \epsilon_{i}\)

Finally, in order to make regression results more interperable in the from of a percent, a log-lin equation was introduced. This means that violent crimes per thousand would be logged, producing coefficients that can then be read as percents. This provides a better general feel for results. The final regression was represented by the equation:

\(Log(Violent CrimesperThousand)_{i} = \beta_{0} + \beta_{1}GunLaws_{i} + \beta_{1}PovertyRate_{i} + \beta_{1}Climate_{i} + \beta_{1}GDPperCap_{i} + \epsilon_{i}\)

Data

Data was collected from individual states from the years 1997-2014. In the final dataset, there were 900 observations of 7 different variables, which included the year and state variable and the 5 variables mentioned above.

Data on violent crime was pulled from the Uniform Crime Reporting Statistics website, a database put together by the FBI as they gather statistics from law enforcement agencies across the nation. The data gathered includes variables accounting for totality of violent crime, crime rates per 100,000, the population, and subsections of specific crimes commited (i.e burglary, theft, larson). Violent crime was selected out of this, as the goal was to look into the totality of crime compared to any specific type.

Data on poverty rates was collected from the United States Census Bureau were it was presented in table form containing each state’s population, total people in poverty, and a ratio of population/poverty. Population was trimmed from this and only the states total number of people in poverty were used going forward. Poverty was then converted to a rate, dividing the total number of people in poverty by the population.

GDP data was collected from the Bureau of Economic Analysis. The BEA is from the United States Department of Commerce, a government agency. They provide macroeconomic and industry statistics of the country.

Climate data was pulled from the website https://www.currentresults.com/Weather/US/average-annual-state-temperatures.php. Here they utilized data from the NOAA National Climatic Data Center. The temperatures provided were averages from each state during the years 1971 to 2000.

Key Measures

The dependent variable and the measurement examined in this experiment was crime rate. This variable accounts for the total number of crimes committed in a single state in one year, divided by the population of that state in that year times 1000. This is then the violent crime rate per 1000 residents of that state.

The independent variable of focus is the Gun variable consisting of four levels. These levels are described above. The coefficients of these four factors in our regression will be important in drawing conclusions on the effectiveness of open carry laws on preventing violent crimes.

Other independent variables included poverty rate (consisting of the total citizens in poverty in the given state divided by the population), GDP per capita, and weather (consisting of average temperature of the state from years 1971-2000).

Variable Measurement Type Interest Variable Measurement Type Interest
Crime Rate Violent crimes committed per 1000 people in population Rate Dependent variable Poverty Rate Total people in poverty divided by total population Rate Independent variable
Gun Number 1-4 (1 = No permit open carry, 2 = open carry with permit, 3 = open carry with permit and regulations, 4 = no open carry) Factor Independent Variable Population Population of State measured in thousands Number Independent Variable
Variable Measurement Type Interest
GDP per Cap GDP divided by population Ratio Independent Variable
Weather Temperature measured in Fahrenheit Number Independent Variable

Descriptive Statistics

Variable Maximum Minimum Mean Standard Deviation
Crime Rate (per 1000 population) 10.236 0.669 4.038 1.783
Poverty Rate 0.231 0.045 0.125 0.034
Weather 70.7 26.6 51.944 8.624
Poverty Rate 79.863 21.362 41.853 10.744

Looking at the varying gun laws mapped against the crime rates, there is no apparant pattern that jumps out. The case may be made that Gun Level 4, no open carry allowed, appears relatively higher than other levels.

Results and Robustness Checks

Dependent variable:
Crime.Per.Thousand
(1) (2)
Gun2 0.030 -0.065
(0.134) (0.131)
Gun3 0.117 -0.014
(0.213) (0.199)
Gun4 1.794*** 1.317***
(0.180) (0.176)
Poverty.Rate 4.733**
(1.845)
GDP.cap 0.007
(0.005)
Weather 0.067***
(0.007)
Constant 3.805*** -0.450
(0.077) (0.447)
Observations 900 900
R2 0.105 0.231
Adjusted R2 0.102 0.226
Residual Std. Error 1.690 (df = 896) 1.569 (df = 893)
F Statistic 35.067*** (df = 3; 896) 44.740*** (df = 6; 893)
Note: p<0.1; p<0.05; p<0.01

The first linear model consisted of predicting crime per thousand by the factors of gun laws. This simple regression of 900 observations yielded an Adjusted R2 of 0.102. Gun level 2 and 3 return insignificant, while level 4 is significant.

The second linear model consisted of predicting crime per thousand by the gun laws, poverty rate, GDP, and climate. When adding the variables, it shows that the gun levels were positively biased, as their coefficients have dropped. This gives evidence that the added variables provide insight, as it shows that there was bias present.

Dependent variable:
log(Crime.Per.Thousand)
Gun2 -0.018
(0.035)
Gun3 0.068
(0.053)
Gun4 0.289***
(0.047)
Poverty.Rate 1.563***
(0.490)
GDP.cap 0.006***
(0.001)
Weather 0.021***
(0.002)
Constant -0.263**
(0.119)
Observations 900
R2 0.258
Adjusted R2 0.253
Residual Std. Error 0.416 (df = 893)
F Statistic 51.736*** (df = 6; 893)
Note: p<0.1; p<0.05; p<0.01

Moving to a log-lin regressions provides easier results to interpet. This regression says that a state in which allows open carry with a permit would expect a decrease in crime per thousand of 1.8%. This result again comes back not statistically significant. When comparing the change of crime rate given in the coefficient to the standard deviation of crime rate, it shows that this is not a drastic change, and can be considered not practically significant.

A state which allows open carry with a permit and restrictions would expect an increase in crimes per thousand of 6.8% compared to a state that allows open carry with no permit nor restrictions. This coefficient also returns statistically and practically insignificant.

States which ban the practice of open carry would expect an increase of crime per thousand of 28.9%. This coefficient came back statistically significant. The value of the coefficient is also relatively large, so it can be deemed practically significant. A 28.9% increase in crimes per thousand can lead to the increase of thousands of crimes.

Every one of the added outside independent variable return significant results. However, GDP and weather produce relatively small coefficients. Poverty rate seems to be a very good indicator of violent crimes.

We can not draw conclusions as to determine that the varying factors of gun legislation cause this difference in violent crime. It may be expected that a state that completely bans open carry, such as California, may have an initially high crime rate. Therefore, legislation might want to ban the practice of open carry to prevent that rate from increasing. Introducing this legislation may not alter the crime rate at all, and California may just have a high crime rate no matter what laws are passed.

Conclusion

After eliminating potential omitted variable bias and running a semi-log regression on the acrquired data and desired variable to make coefficients easier to interpret, it appears that there is not a significant difference between states that allow open carry without a permit, states that allow open carry with a permit, and states that allow open carry with a permit and restrictions. However, states that do not allow open carry at all do expect to see a significant different in violent crimes, both practically and significantly.

These findings do not prove any causality. It can not be concluded that the different levels of open carry legislation contribute to differing levels of violent crime. This experiment paved the way for a difference in difference model, which can be utilized to analyze pre and post trends. This model will prove useful as it allows for the treatment to occur at different times. After this regression has been ran, results may lead towards a stronger conclusion of causality. Due to time restraint, the inability to find state legislation dates, and small sample size of levels 3 and 4, this difference and difference model was unable to be explored. This experiment, however, lays out the path and can be replicated to then calculate this regression.

Works Cited

National Institute of Justice. (2019). Gun Violence in America. Retrieved from https://nij.ojp.gov/topics/articles/gun-violence-america

National Press Office. (2019). FBI Releases 2018 Crime Statistics. Retrieved from https://www.fbi.gov/news/pressrel/press-releases/fbi-releases-2018-crime-statistics

Open Carry. (n.d.) Retrieved from https://www.usconcealedcarry.com/resources/terminology/carry-types/open-carry/

ProCon.org. (2019). State-by-State Concealed Carry Permit Laws. Retrieved from https://concealedguns.procon.org/state-by-state-concealed-carry-permit-laws/

Gramlich, J., Schaeffer, K. (2019). 7 facts about about guns in the US. Retrieved from pewresearch.org/fact-tank/2019/10/22/facts-about-guns-in-united-states/

Moorhouse, J., Wanner, B. (2006) Does Gun Control Reduce Crime or Does Crime Increase Gun Control?

Data Retrieved From:

https://www.ucrdatatool.gov/

https://www.census.gov/

https://www.currentresults.com/Weather/US/average-annual-state-temperatures.php

https://www.bea.gov/

Map Made On:

https://mapchart.net/usa.html