Don Boudreaux writes on the law and economics of marginal deterrence. The notion isn't just ivory tower theorizing; for example:

In August 2000, Nicholas Markowitz, 15, was kidnapped in a Los Angeles suburb, held for several days, and then murdered by a group of men who had grown up with Markowitz' older brother. The larger motives for this crime remain murky, with allegations of prior drug dealing, drug debts, and attempted insurance fraud on the part of the older men still unresolved. It is clear that the young victim was innocent. What is clear as well is that the de facto absence of a death penalty in California led to the decision to murder the teenager.

After one of alleged killers, Jesse James Hollywood -- no, I am not making that up -- "called his lawyer and learned the severe penalty for kidnapping, police say, the young men decided they had to kill Nicholas" (Los Angeles Times, August 26, 2001). In other words, since the penalty for kidnapping effectively was a life sentence or something close to it, and since the perceived probability that the kidnappers would be apprehended and convicted was high, the marginal (or "extra") penalty for murdering Markowitz was perceived to be low or zero, in that the likelihood of actual subjection to capital punishment in California is trivial -- California has executed ten individuals since 1976. And it's a punishment that, in any event, is delayed one or two decades by the appeals process. Accordingly, the prospective act of murder, even upon arrest and conviction, would yield the same life sentence already looming. So why not get rid of the central witness to the kidnapping?