Is social welfare increased (using Rothbard's demonstrated preference criterion) when I pay a ransom to free a hostage?

From Rothbard's essay on the subject to remind everyone of the details... >Now what happens when the State, or a criminal, uses violence to interfere with exchanges on the market? Suppose that the government prohibits A and B from making an exchange they are willing to make. It is clear that the utilities of both A and B have been lowered, for they are prevented by threat of violence from making an exchange that they otherwise would have made. Clearly we can't say that paying a ransom is one's demonstrated preference, because it is an action motivated by coercion. So how do you argue that individuals should be allowed to pay ransoms? If you say "it's their utility-maximising choice under the circumstances", I could easily reply that paying one's taxes similarly increases utility. I could say it is a Pareto improvement to let people pay ransoms, but not sure how to put it into Rothbard's language. I don't think he would say social welfare is increased by government allowing individuals to pay ransoms, because all it is doing is enabling B to rob A. But then A would prefer to be robbed than see C be killed...

link: http://bit.ly/1ISr0uz