As a corollary to the proposition that all institutions must be subordinated to the law of equal freedom, we cannot choose but admit the right of the citizen to adopt a condition of voluntary outlawry. If every man has freedom to do all that he wills, provided he infringes not the equal freedom of any other man, then he is free to drop connection with the state—to relinquish its protection, and to refuse paying towards its support.
If I don't want a government service, why can't I stop paying for it if I stop consuming it? In a phrase, why can't I opt out? It's a question libertarians like Herbert Spencer have been asking for quite a while, and it's got a lot of intuitive appeal. If I don't send my kids to public school, why can't I get a refund of the fraction of my property taxes devoted to education?
Economists usually view this question as the height of naivete. "Of course you can't opt out! Government provision wouldn't work if people could stop paying taxes for services they say they don't want."
On reflection, though, there are at least three distinct reasons why opting out would kill government provision:
1. Strategic reasons. Due to externalities, people continue to enjoy most or all of the benefits of the government service after they officially opt out. So there would be an incentive to opt out even if you did prefer (having the good + paying the associated taxes) to (not having the good + not paying the associated taxes).
2. Redistribution. Some people are paying more than they are getting; others are getting more than they are paying. So there would be an incentive for all net losers to opt out, making redistribution unsustainable.
3. Wasteful spending. People genuinely value the good at less than its cost. End of story.
Now notice that only (1) has anything to do with economic efficiency. If opting out is purely strategic, then opting out leads government to stop providing goods that people value at more than their cost. The honest libertarian could still say that liberty is more important than economic efficiency, but he'd have to admit that there's a conflict.
However, in cases (2) and (3), matters are different.
In case (2), opting out would make redistribution unsustainable. But that's not an efficiency problem. In fact, since redistribution almost always has some deadweight costs, letting people opt out would increase efficiency. The libertarian policy leads to marginally greater economic efficiency.
In case (3), the situation is even more stark. If people opt out simply because people value a good at less than its cost, then economic efficiency demands opting out by definition. The libertarian policy is required for economic efficiency.
During my time in economics, I've noticed is that mainstream economists are very quick to assume that all opting out is strategic. But that's not true even in the "clearest" cases. Take national defense. These days, many leftists would clearly like to opt out - and put scare quotes around "defense"! We can fairly say that most of this is NOT strategic. Not only do they value e.g. U.S. policy in Iraq at less than its cost; they would probably pay to reverse U.S. policy in Iraq.
And if that's true for national defense, it's even more clearly true for things like public education. We can pretend that opting out is strategic. But the real reasons are that's it's redistributive (from families that don't use the public schools to those that do), and wasteful (even families that do use the schools often value the services at less than the cost in taxes).
Bottom line: Libertarians should admit that there are special circumstances where opting out leads to economically inefficient outcomes. But economists should admit that these are special circumstances that actual government spending often fails to satisfy. The economist who points out the splinter of naivete in the libertarian's eye ought to look at the beam of naivete in his own.