Sunday, June 24, 2012

Reciprocal Fear and the Castle Doctrine Laws

In his timeless classic The Strategy of Conflict, Thomas Schelling began a chapter on the "reciprocal fear of surprise attack" as follows:
If I go downstairs to investigate a noise at night, with a gun in my hand, and find myself face to face with a burglar who has a gun in his hand, there is a danger of an outcome that neither of us desires. Even if he prefers to just leave quietly, and I wish him to, there is danger that he may think I want to shoot, and shoot first. Worse, there is danger that he may think that I think he wants to shoot. Or he may think that I think he thinks I want to shoot. And so on. "Self-Defense" is ambiguous, when one is only trying to preclude being shot in self-defense.
This effect is empirically important, and is part of the reason why homicide rates vary so greatly across otherwise similar locations, and can change so sharply over time at a given location. In our attempt to understand why the Newark homicide rate doubled in just six years from 2000-2006 while the national rate remained essentially constant, Dan O'Flaherty and I found a substantial number of homicides to be the outcome of escalating disputes between strangers or acquaintances often over seemingly trivial matters. High rates of homicide make for a tense and fearful environment within which the preemptive motive for killing starts to loom large, and this itself reinforces the cycle of tension, fear, and continued killing. Incremental reductions in homicide under such circumstances are unlikely to be feasible, but sudden large scale reductions that transform the environment and break the cycle can sometimes be attained. Similar effects arise with international arms races.

In the jargon of economics, homicide is characterized by strategic complementarity: any increase in the willingness of one set of individuals to kill will be amplified by increases in the willingness of others to kill preemptively, and so on, in an expectations driven cascade. Any change in fundamentals can set this process off, such as a breakdown in law enforcement, easier availability of firearms, or increases in the value of a contested resource.

The logic of strategic complementarity implies that a broadening of the notion of justifiable homicide, in an attempt to benefit potential victims of crime, can have tragic and entirely counterproductive effects. Florida's 2005 stand-your-ground law is an example of this, and more than twenty other states have adopted similar legislation in its wake. These are sometimes called castle doctrine laws, since they extend to other locations the principle that one does not have a duty to retreat in one's own home (or "castle").

Enough time has elapsed since the passage of these laws for an empirical analysis of their effects to be be conducted, and a recent paper by Cheng and Hoekstra does exactly this. Determining the causal effects of any change in the legal environment is always a tricky business. The authors tackle the problem by grouping states into those that adopted such laws and those that did not, and comparing within-state changes in outcomes across the two groups of states (the so-called difference in differences identification strategy). Their findings are striking:
Results indicate that the prospect of facing additional self-defense does not deter crime. Specifically, we find no evidence of deterrence effects on burglary, robbery, or aggravated assault. Moreover, our estimates are sufficiently precise as to rule out meaningful deterrence effects. 
In contrast, we find significant evidence that the laws increase homicides... the laws increase murder and manslaughter by a statistically significant 7 to 9 percent, which translates into an additional 500 to 700 homicides per year nationally across the states that adopted castle doctrine. Thus, by lowering the expected costs associated with using lethal force, castle doctrine laws induce more of it... murder alone is increased by a statistically significant 6 to 11 percent. This is important because murder excludes non-negligent manslaughter classifications that one might think are used more frequently in self-defense cases. But regardless of how one interprets increases from various classifications, it is clear that the primary effect of strengthening self-defense law is to increase homicide.
These are statistical findings and refer to aggregate effects; no individual homicide can be attributed with certainty to a change in the legal environment, not even the one killing that has brought castle doctrine laws into national focus. Nevertheless, we now have compelling evidence that the adoption of such laws has led directly to several hundred deaths annually nationwide, with negligible deterrence effects on other crimes. While the latter finding may be surprising, the former should have been entirely predictable.


  1. This sounds very much like a "fooled by randomness" issue. Would you not expect some locals to be outliers? Does there have to be a site specific reason?

    Sociologists who asked armed robbers (muggers) what their greatest fear was, and they said that it was armed civlians - more so than robbing an police officer out of uniform. Of course, since they are still armed robbers, it may not have a statistical impact. Or the impact could be washed out by a perception that people who are likely to have guns (drug dealers, jewelry store managers, etc), are also the most likely to have cash.

  2. You expect outliers in both directions... too many outliers in one direction suggests that this is not randomness. Look at the paper for details but they are very clear about this. See also

    which finds similar results. Your point about the fear of armed civilians actualy supports the argument made here, since fear is the main driver of preventive killing.

  3. Self-Defense is NOT murder. Anyone in my home at two in the morning - uninvited- is not there to be a Good Samaritan.

  4. Hiflyer, that's right, killing in self-defense is not classified as murder or non-negligent manslaughter, it is classified as justifiable homicide. That's exactly why this post focused on murder and manslaughter, not homicide in general.

    If you look at the paper, you'll see that justifiable homicides also rose in states with stand-your-ground laws but the effect was not statistically significant. The really significant effect was on killings that were not of the kind you describe. I suspect that many of these were preemptive (but not premeditated) killings arising from escalating disputes.