I'm not sure if your objection is that our economy should put nonzero resources into first-world education. But if it is, note that the issue is where the next marginal $5000 should go — whether, given the current allocation of resources, marginal dollars do more good in first-world education than in third-world health. If our hypothetical teacher/trader's donation is somehow matched by our entire economy, or even by all philanthropists, the AMF's room for more funding would be instantly exhausted.
If you want to figure out the value of an education in the first world, you would have to take the cost of the education and compare it to the amount of added value that that person will produce over the course of his lifetime because of that education.
Agreed, and on the other side one should likewise add up all the value that two saved children in Africa would produce over their lifetimes. As I said, I'm not sure I agree that the education intervention produces more total value, but perhaps I could be persuaded of this.
But if it is, note that the issue is where the next marginal $5000 should go — whether, given the current allocation of resources, marginal dollars do more good in first-world education than in third-world health.
I don't think that's the most important issue here. If teacher A is creating more value at teaching than he would be as a bond salesman donating to charity, then the government could (in theory) then shift $5000 worth of resources from education over to foreign aid as the teacher is generating more then $5000 worth of extra education value. ...
A very interesting article on "earning to give", featuring LessWrong members Jeff Kaufman, Julia Wise, Holden Karnofsky, William MacAskill and Toby Ord. Some excerpts: