A game theoretic model of gun control

Abstract

With the passage of the Brady Bill in 1993 and the approval of the Feinstein Amendment the following year, gun control has once again become a subject of heated public debate in the United States. Procontrol forces point to high levels of gun ownership as the cause of much, if not most, violence in America, and argue that gun control will inevitably reduce violence by reducing the number of guns in circulation. Anticontrol forces respond that high levels of gun ownership are a response to, not a cause of, violent crime, and that gun control disarms law-abiding citizens, not criminals. Certain facts are clear. Approximately half of all American households are armed, giving the U.S. the highest rate of gun ownership in the world. Switzerland is probably second, with one-third of its households armed (Kleck 1991, 19). The relationship between levels of gun ownership and violence is harder to discern, however, as is the connection between gun control and violence. A large number of empirical assessments of gun control legislation have appeared over the past two decades.’ A recent article by Kleck and Patterson (1993) gives a useful overview of this literature. Kleck and Patterson note that the overwhelming majority of the studies either were inconclusive or found that gun control laws had no significant impact on violent crime. They conclude their literature review with the following statement: “taking prior research as a whole, it would be fair to say at this point that a consistent, credible case for gun control efficacy in reducing violence has not yet been made.” Given that the empirical literature has largely failed to resolve the gun control controversy, researchers might be well advised to take a more theoretical approach to the problem. The construction of relatively simple models of victim-criminal interaction might provide interesting insights about the desirability of gun control legislation and the mechanisms of its influence. Rarely in the gun control debate has either side stated its assumptions clearly. Formal modeling would require these as-