Gorton’s triple-A error

By Felix Salmon
March 22, 2010
official paper for the US Financial Crisis Inquiry Commission. Here they are, in situ:

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

Beware of academics wielding exclamation marks!! Gary Gorton is a very highly respected professor in the fields of finance and economics, but that doesn’t stop him throwing double-shrieks into his official paper for the US Financial Crisis Inquiry Commission. Here they are, in situ:


The problem here is that while double-A-rated corporate bonds did indeed trade through triple-A-rate corporate bonds at the end of 2008 and beginning of 2009, Gorton’s explanation — even with two exclamation marks attached — is entirely wrong.

Finance Guy gives the long version of the takedown, and it’s well worth reading. But the short version is simple. Gorton thinks that triple-A-rated corporates gapped out because they were being sold off in “fire sales”. But in fact, triple-A-rated corporates in general didn’t gap out at all. General Electric gapped out, for very good reason: as David Merkel recalls, “GE Capital nearly bought the farm in early 2009″ due to the fact that it had a major maturity mismatch and was having difficulty rolling over the short-term liabilities with which it was funding its long-term assets.

In March 2009, GE’s CDS were trading at a spread of more than 1,000bp – and GE’s bonds made up the majority of the index of triple-A commercial bonds. (There are precious few triple-A commercial credits these days, and most of them issue very little in the way of bonds.) So it’s hardly surprising that the triple-A commercial-bond index gapped out — and the reason for it has nothing to do with forced selling, or even with selling at all. Indeed, my guess is that almost no GE bonds were sold during those periods.

If Gorton wants to provide evidence of forced sales of high-rated corporate debt at certain periods of time, he’s going to have to provide some volume figures, rather than trying to extrapolate volumes from price charts. Because the chart he provides simply doesn’t show what he says it shows.

(Incidentally, this is yet another example of the blogosphere being extremely good at fact-checking claims by experts. When done well, everybody wins: see for instance my friend Stefan Geens’s refutation of an article by the economist Craig J. Richardson entitled Visual Evidence of the Cost of Destroying Property Rights which was then picked up by Alex Tabarrok of Marginal Revolution. Once Alex saw Stefan’s post, he prominently updated his post to reflect the newly-revealed facts of the matter. I wonder whether Gorton will do anything similar.)

More From Felix Salmon
Post Felix
The Piketty pessimist
The most expensive lottery ticket in the world
The problems of HFT, Joe Stiglitz edition
Private equity math, Nuveen edition
Five explanations for Greece’s bond yield
6 comments so far

Multiple exclamation marks. A sure sign of a diseased mind. :D

Posted by owe.jessen | Report as abusive

1. The chart purports to show spreads between AAA and AA bonds, not the spread between AAA and AA CDS’s.

2. How can you calculate a spread if there are no trades? There should be gaps in the line where trades were not occurring.

3. Yahoo finance says that at Mar 31 09 GE had $176 billion of s/t debt outstanding and $332 billion of LT debt o/s. It is your suggestion that NONE of it traded during the greatest financial crisis of the past 50 years. Well ho-kay then!

4. If I read it correctly, what you are trying to say is that if you take out the largest issuer of AAA paper, AAA spreads would be different. Good point! By the way, if you exclude the most indebted members of the Eurozone, the Eurozone is highly solvent.

Posted by johnhhaskell | Report as abusive

“2. How can you calculate a spread if there are no trades?”

Dealer quotes

Posted by GingerYellow | Report as abusive

I don’t have good TRACE data, but if you look at the exchange-traded GE bonds (GEA, GEG), volume spiked at each of the crisis points in 2008-2009. I’m quite certain that there was significant volume in standard GE bonds, but someone else will have to check out the numbers.

The main problem with Gorton’s thesis that AAA-AA spreads indicate fire sales is that it makes no sense. If AA are trading higher than AAA, ratings are accurate, and you want to raise cash, why wouldn’t you sell the AA instead? AAA/AA spreads are meaningless when the ratings aren’t accurate. I assume the huge spike back to positive territory in 2009 comes from the GE downgrade. Thus what Gorton gives you is not a chart of credit market behavior, but a chart of rating agency behavior.

The other complicating factor in this analysis is that most of the actual AAA issuance in late 2008/early 2009 was government-backed, since private securitization was dead and GE certainly didn’t want to issue debt at 10%+. So these spreads don’t tell you anything about access to capital, since the TBTF companies were issuing huge amounts of government-backed debt, receiving huge government capital infusions, and continuing to securitize assets through Fannie and Freddie.

Posted by najdorf | Report as abusive

congratulations najdorf, you figured it out.

As for GingerYellow, please note that the volume of GE debt outstanding is greater than the amount of Greek sovereign debt outstanding. No trades during the crisis? Yeah right.

Posted by johnhhaskell | Report as abusive

Eh? I never said there were no trades. I just pointed out that to the first commenter there are ways to obtain spreads even if there are no trades – although obviously they aren’t as good indicators of “true” value as real trades. For that matter, even in the absence of dealer quotes, you could have indicative bid/offer levels.

Posted by GingerYellow | Report as abusive
Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/