AIG priced a giant stock offering last night at $32.50, making the government rich. A really really simple question you could ask about AIG is “how’s the government doing?” and I Googled around for the answer yesterday and got increasingly frustrated, then angry, then drunk. Why can’t someone tell me that? The answer has to do I think with competing interests and secrecy and embarrassment and innumeracy both real and tactical, and I could write a book about it but won’t.1 Instead, I will just tell you how the government is doing on AIG, and then you will know.2
- The government has gotten back
$12.3$15 billion more than it put into AIG so far, plus it has about $10bn$8 billion worth of AIG shares left over. (This is what the government says too, to within rounding error.) [Update: revised for greenshoe exercise.]
- So, great!
- Its IRR is
3.2%3.9%, or 5.7% if you assume it sold the remaining AIG shares today (which: it didn’t).
- If you assume the government’s cost of capital for the bailout was 3.04%, or roughly 5-year Treasury rates as of the time they signed on to this almost 5-year bailout, then the government’s made an economic profit (returns in excess of cost of capital) of
$600 million$3 billion, or $9.9 billion including the remaining AIG shares.
- If you assume the appropriate discount rate for the bailout was 12%, or roughly what AIG’s initial Fed credit facility paid, then the government has undercharged AIG by about
$26.3$24.6 billion, or $19.7 billion including the remaining shares.
- Neither is a good assumption.3
- The end.
Here is one speculation: Hank Greenberg? The government is more or less out with what looks like a pretty modest profit for saving his company. How does that influence whether you think he should win his lawsuit saying that the government unfairly diluted him out of a perfectly healthy AIG? Some people think there’s something to that suit, but to really feel for him you need to think that the government got a really sweet deal on its bailout. On the evidence so far – which is basically all of it – that doesn’t seem especially true.
A Horrible Spreadsheet [Google Docs]
Treasury sells big chunk of AIG stock at a profit [Reuters]
Plot Twist in the A.I.G. Bailout: It Actually Worked [DealBook]
$182 Billion Commitment to AIG During Financial Crisis Now Fully Recovered After Pricing of $18.0 Billion Treasury Common Stock Sale [EDGAR]
Overall $182 Billion Committed to Stabilize AIG During the Financial Crisis is Now Fully Recovered [Treasury]
1. As of this morning the situation is better, with a nice overview accounting from the government in the AIG press release itself [update: and this very nice set of pictures from Treasury]. Some other useful efforts include this paper (no math but lots of bailout details), this GAO report, this ProPublica accounting (Treasury only), this Treasury accounting (Treasury only), and the Fed’s disclosure (Fed only and … diffuse). Unsatisfactory efforts include calculations of “breakeven prices,” all kinds, and SIGTARP’s weird efforts to partition TARP from non-TARP and bang on about TARP losses even when they’re Treasury gains.
2. Also if you know someone who’s already figured this out and put it online in a user-friendly way:
- don’t tell me, because I wasted a lot of time and anger on this and am not sure I could take it, and
- tell the person who’s already figured it out to improve his/her search engine optimization. My inbox is full of promising SEO leads if s/he needs a recommendation.
3. However: I like discounting at 12% because, at the moment that the government committed to the bailout, that’s what it was charging on its 2-year loan commitment. Plus 80% of the equity but whatever. I don’t think anyone at the time thought 12% was too high, except maybe Hank Greenberg and probably not even him. If you’re discounting these cash flows at less than 12% I think you need a reason. (The fact that some of them are collateralized, etc., is a plausible one.)
4. If you think I’m wrong read this note then email me. But read the note first.
Basic methodological question number 1: What is the government? I started with “programs special to AIG,” meaning the Fed credit facility and a few related lines, the Treasury bailout that restructured it, and Maiden Lanes II and III. So not, in particular, the commercial paper program, which AIG partook of. But also not, like, low fed funds rates or whatever. I don’t think that those things are as bailout-y as the specific AIG lashings of money. You disagree, that’s fine. I also don’t count AIG’s ability to deduct losses for tax purposes as a bailout – again, not special to AIG – despite the fact that the bailout changed rules that would have wiped out those losses without the rule changes. That doesn’t sound like extra bailout to me, but if it does to you – and it does to Andrew Ross Sorkin [update: or did in February; this is a bracing riposte from the broad-view-of-bailouts school of thought] – then have at it.
Basic methodological question number 2: Commitments or cash flows? My aim here was to measure actual disbursements of cash by the Fed or Treasury, not undrawn commitments. As a matter of credit risk this is aggressive – the government was on the hook for all committed amounts – but as a matter of calculating IRRs it seemed easier. I am somewhat ashamed of this choice.
One other source of shame is that I am not rigorous about “maximum amounts outstanding” – when one program replaces another I have two separate line items, which means that when the Fed says its maximum commitment was $112.5bn and I have a total of $172bn for them, it comes from combining maxima for each separate item. Theirs is philosophically better but, again, mine makes it easier for my feeble brain & spreadsheet to calculate IRRs.
Ooh, more shame: many items are glommed into the last months of quarters (because disclosed in quarterly reports) rather than digging through (often non-existent) contemporaneous sources to find exactly when they happened; this makes IRR and NPV calcs necessarily imprecise but cash flows are accurate to within a quarter.
Finally, sourcing is I think adequately explained in Column BE of the “Horrible Math” tab of the spreadsheet. The spreadsheet is intended to be … um … not quite intuitive, but close.