Adam Schaeffer of the Cato Institute manages the fairly impressive feat of beating up on me in a blog post (I am an "ingrate," along with Matt Yglesias and Felix Salmon) without mentioning or responding to the argument I was actually making. So hey, let me just go ahead and make it again: The tax code's definition of a charity is too broad. Do you disagree, Adam?
Still, the separate question Schlaeffer asks is an interesting one: "Why shouldn't we charge rich parents tuition to attend public schools? If a charitable deduction for private schools is so bad, why isn't a free public education even worse?" So let me take a crack at it.
In general, means testing public services is a good idea, for obvious reasons. A concept of justice that said, "everyone gets the same amount, regardless of how much they need or deserve" would not be a very convincing concept indeed. But the case for means testing is often overstated and sometimes treated as gospel. And I can think of five reasons why it shouldn't be:
1. Means testing creates some inefficiencies. Deciding to make some government service contingent on one's demonstrated need requires creating some administrative mechanism for measuring need. This is costly and difficult. It is also (as I'm sure I don't need to tell the good people of Cato) potentially incentive-distorting: limiting a public service to people below a certain income should, at the margin, reduce the incentive to rise above that income.
2. Means testing can reduce the political support for a service. This isn't that complicated. If a service is provided to everyone in the United States, then everyone in the United States has some incentive to support it. If a service is provided to only the neediest individuals, then not everyone has an incentive to support it. (As an historical matter, my understanding is that this is one reason why social security benefits are not means tested.)
3. Means testing can reduce the quality of a service. This is related to #2. If not everyone in a given community benefits from a service, then the community will be less inclined to offer a robust form of that service. (And I think both #2 and #3 are exacerbated by the fact that the individuals most in need of public services tend to have the least political power.)
4. Means testing can create stigma, since it identifies some people as needy. (That might sound a bit hoity-toity, but I think it would be important in the context of a public education.)
5. Means testing can be immoral. Some benefits of citizenship are thought of as rights -- things to which we are entitled regardless of expected benefit or demonstrated need. No one, for instance, has recently suggested that voting should be means-tested.