1) On balance, I'm sympathetic to the DeLong analysis of the bailout plan. The key point of Krugman's criticism is that he assumes we must be in one of two states: either the assets are undervalued at current marks or they're not. And this just isn't the case. Certainly, at least some of the assets are undervalued at current marks -- maybe not all, but just by assuming there's some distribution you can reasonably think that some are. If you further think that none of the usual bond suspects (banks, insurance companies) in their right mind want to own up to their investors that they hold subprime assets at any price or even lend against them, you can imagine that the only bids you'll be getting are at levels that provide equity-capital-driven speculators 20% (maybe higher, crisis) return targets unlevered (since there is no leverage to be had). So I think it's far from inconceivable that a fair chunk of these assets are priced underneath some kind of fundamental value.He follows up later in another email:
2) I still have no real basis on which to assess Geithner's performance. It remains the case that if he were actually out to nationalize big banks he would be behaving no differently than he is now -- making happy noises, giving little inklings of his plans that don't screw over equity necessarily, and so on, all in preparation for the Night of the Long Tier 1 Capital Ratios that will be nationalization. At the same time, sadness that none of it is actually happening yet, that all existing plans appear to suck, etc.
3) Trying to nationalize B of A (say) without a damn good reason is probably inviting the legal showdown of the century. If the bank regulator acted not in accordance with existing regulations or did any little thing wrong you can imagine a lawsuit for $10b or even $100b very easily. Unless you get actual cooperation from management (Bear, AIG did cooperate, remember) for whatever plan you have, you're going to need an airtight case -- arbitrary action will probably stick you with a preliminary injunction and a giant court fight. You may remember from a recent NY times story that banks are marking at 60 securities that are currently selling for 30, on the grounds that the markets are too illiquid and firesale-based to trust; the only reason this situation should persist is that it can get by FDIC accountants, at least scared of biggest-lawsuit-in-history FDIC accountants. In keeping with this point, there's another virtue of TALF: you do get fair market bids on essentially everything in the universe, and bids that don't reflect distressed sales. If we're in the Krugman Scenario, where even at 6:1 leverage prices the big banks are insolvent, we have pretty incontrovertible evidence of that fact, which we can then use to safely put said banks down.
Congressional action probably can substitute for the implicit FDIC-accountant-stick in (3) (that is, managements can be made to cooperate either through the threat of old-style regulatory takeover or some new-style legal action); I'd probably still prefer to do without and avoid the congressional wrangling, but it's doable.
Many of the numerical examples floating around the internet are implicitly structured to maximize the value of the Geithner put. The "assets worth zero or 100" example is probably the worst offender because this is exactly where Brad DeLong's point bites: if senior tranches of pools of prime first-lien mortgages are worth zero, not only has essentially every borrower in the country defaulted but the houses securing those mortgages are also not worth the price of selling them, i.e. nobody has any money at all and it really is canned-goods-and-bottled-water time. On the other end of the scale you could use a binary model where assets were worth either 15% more or 15% less than new purchase price -- in this world, investors make +100% or -100% and the Geithner put is valueless (since the assets can't go below the value of the loan). Probably we're somewhere in between, but the point is that choice of constants really matters here and some of the more popular choices are wildly misleading.
I should further point out that, while I think that recourse loans would generate many of the same effects, I can see the case for making them non-recourse, particularly if you want to raise dedicated new money and/or don't want to do extensive credit analyses on every possible investor, but that's something I'll have to expand on when not running for a plane.