Yes, I disagree with the "obvious" conclusion that college graduates are most valuable. Additionally, I think the argument put forward supposedly in support of that conclusion - i.e. that we've invested more in graduates - actually leads to the exact opposite conclusion. As others have pointed out, it's not valid economics but pure sunk-cost fallacy.
I think all he's getting at is recent college grads have the highest Net Present Value. It's not a matter of sunk cost.