David made me think a second time about second-best immigration policy.  My thoughts on his:

1. I think Bryan drastically understates the ability of even fairly
low-skilled workers to come up with a substantial five-figure admission
fee. (I’m assuming the fee would be somewhere between $40K and $100K.)
My basis: look at what many of these same people are willing to pay “coyotes”
to smuggle them here. And remember that this is a situation in which
the people are not assured of making it here and take a substantial
risk of dying.

Fair enough, but look carefully at David’s coyote link.  Countries where you have to pay $10k+ to get to the U.S. send few immigrants here.  The large majority of illegal immigration to the U.S. comes from Mexican, where coyotes charge roughly two years’ income for transport.  Not ten years’ income, or thirty years’.

My view is that even if
the government charged a lump sum of $100K, you would have 5 million
instant takers. If I’m right, that would be a cool $500 billion. Think
of the one-time drop in the deficit.

That’s completely reasonable.  But 5 million is only a tiny minority of the total number of people who want to immigrate.

2. Let’s say Bryan’s argument about the inability of people,
especially the unskilled, to raise that kind of cash is right. That
doesn’t imply a 10-percentage-point increase in the income tax rate. It
implies installment payments. If the $100K is the right amount
(remember, I’m in a second-best world here: Bryan would say the right
amount is $0), then the government could charge the person $10,000 a
year for 15 or 16 years. That avoids the substantial dead-weight loss
that comes with higher marginal tax rates.

I completely agree about the deadweight cost.  And on reflection, if the surtax goes away after the immigrant pays off a fixed sum, my proposal and David’s are nearly equivalent.  My only concern is that I think an “immigration tax” would have marginally more popular appeal than an “installment plan.”  But I could be wrong.