Via Christopher Shea, this paper claims to show that “castle doctrine” laws actually increase murder rates. Abstract:

Since Florida adopted the first castle doctrine law in 2005, more than 20 other states have passed similar self-defense laws that justify the use of deadly force in a wider set of circumstances. Elements of these laws include removing the duty to retreat in places outside of one’s home, adding a presumption of reasonable belief of imminent harm necessitating a lethal response, and removing civil liability for those acting under the law. This paper examines whether aiding self-defense in this way deters crime or, alternatively, escalates violence. To do so, we apply a difference-in-differences research design by exploiting the within-state variation in law adoption. We find no evidence of deterrence; burglary, robbery, and aggravated assault are unaffected by the laws. On the other hand, we find that murder and non-negligent manslaughter are increased by 7 to 9 percent. This could represent either increased use of lethal force in self-defense situations, or the escalation of violence in otherwise non-lethal situations. Regardless, the results indicate that a primary consequence of strengthening self-defense law is increased homicide.

Unfortunately I can’t find an ungated copy anywhere on the web.