I guess I regret asking yesterday “how much did the government make on AIG?” because you can quibble with that question both on arithmetic grounds – should you count other programs as part of AIG’s bailout, rather than just the bailout-y programs I counted? – and on philosophical grounds. The philosophical objection is something like “the government + central banks sort of run a financial system as a squishy whole, and if you cut out one part of it and measure it you are doing something pretty silly because the rest of it is slowly oozing out around you and suffocating your accounting methods.” Something like that.1
But here is perhaps a fairer question: “how much did AIG pay for the capital that the Fed and Treasury invested in it?” This is a better question because AIG is not as polymorphously oozing as a Financial System; unlike Treasury + Fed it has to report GAAP financials. Also because asking about the cost of various transactions makes it feel better to treat those transactions in isolation from other transactions, and to compare them to transactions that other people did.
Best of all, though, it’s the same question: AIG’s “cost of capital” for the government programs is just the internal rate of return on the cash flows on those programs.2 By my math yesterday – which ties out broadly enough with the government’s math that I’m going to declare it “correct” – the answer is about 5.7% per annum, with the Fed programs costing 7.9% and the Treasury’s programs costing 2.8%.3
So how does that stack up? I dunno, but here are some thoughts. First, the obvious comparison: big-bank TARP was a lot more expensive for the banks. This much more expensive:4
So AIG’s cost-of-bailout was much lower than any of the banks’ cost, despite the fact that its bailout was the biggest and it was arguably the closest to death’s door. (Maybe except Citi.) This is a little bit of an unfair comparison because it counts only the most expensive bit of the big-bank rescues and measures it against all of the AIG rescue; arguably Goldman’s bailout-y government funding included some cheaper programs (of the secured-lending, TALF and discount-window varieties) that are somewhat comparable to the secured loans that I’m rolling into the AIG bailout. (Also arguably Goldman’s bailout-y government funding included the AIG bailout.)
But still, if you look at the most comparable bit of the AIG bailout, the comparison is even starker: if you look only at Treasury programs, which consisted of preferred-plus-warrant investments that devolved into common equity investments (and some preferred interests in SPVs holding equity in some specific businesses), then those are comparable to TARP – and they cost AIG under 3% per year. The Fed, which provided only secured loans, drove a harder bargain and got a 7.9% return, which is not bad for secured lending (unless it’s, y’know, lending in September 2008 secured by AIG structured credit assets, but anyway). That in itself is weird – you’d normally charge more for the riskier investment, not less. (Nominally Treasury charged a lot, but one of the core features of the AIG bailout was charging high stated interest/dividend rates, paying the interest/dividends in kind rather than in cash, and then ultimately wiping out the whole instrument before the cash interest was ever paid.)
You could I suppose look at other comparisons. One that I sort of like is that the average cost of four-year debt for a BBB- rated US industrial company, as of September 22, 2008, according to Bloomberg’s FMCI function, happened to be 5.7067% – a basis point or two away from the actual cost for AIG’s actual four-year bailout signed that day. Was AIG as good as a BBB- company? Better? Worse? (Could a BBB- company actually have funded at that rate that day?)
Meh. I don’t have any dramatic conclusions to draw from this. I put it up mostly to propose a corrective to the boring, zero-discount-rate “the government made $X billion on this bailout so That’s Good” narrative. One basic intuition is that in general huge risky investments should cost more (to the investee) than smaller safer investments, so it’s odd that AIG’s ex post cost of capital was so low – lower, for instance, than the stated 12% rate on that first Fed loan. But of course it’s an ex post cost of capital: riskier investments cost more because they’re more likely to default, as AIG did, repeatedly. (Sorry, “restructure.”) Even after repeated restructurings migrated the government down AIG’s capital structure, it still got back a bit more than par, so you can’t complain too much about its so-so returns. But there’s no reason to ignore them, either.
1. Or what? Like, are low interest rates a cost of bailouts? (Are they even good for banks? For AIG?) If so, whom do they cost? “The government”? “The Fed”? “Taxpayers”? “Savers”? Or, unrelated, do you allocate a portion of Volcker Rule-making costs to AIG? Etc. Seems uknowable-ish.
2. Or, like, the negative of it, whatever.
3. That’s assuming what I assumed yesterday, and also assuming that the Treasury dumps its remaining $8-ish billion of AIG stock today at today’s price. Since it’s not doing that, the IRR will vary a bit depending on when it sells that stock and how much it gets for it.
4. Math here, all data stolen directly from ProPublica’s great “find a bailout” database. A quibble here is that I assume warrants are cut off when the government is repaid, even if the government is repaid by warrant auction; this is relevant for the government’s IRR but arguably not for the bank’s cost of capital, since the warrants remain outstanding and so if the stock keeps going up so does their cost. But of course they could’ve bought in the auction so it’s not that big a deal. Numbers, if you’re interested: