Goldman’s CELF-interest

By Felix Salmon
November 29, 2010
The Epicurean Dealmaker has a very smart gloss on the CELF transaction I wrote about on Friday:

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

The Epicurean Dealmaker has a very smart gloss on the CELF transaction I wrote about on Friday:

Investment banks traditionally thought of market making as a client service. An agency business. We put our capital at risk to facilitate the trading of our investing clients. In exchange, we earned a small commission, the occasional chance to put our capital to work in longer-term trades where we thought we had an edge, and—most importantly—priceless insight into the daily operations of particular securities markets, including the appetites, biases, and weaknesses of countless third party market participants. This insight is incredibly valuable, not only in market making itself, but also in making the investment bank possessing it a better informed underwriter for new securities. Securities markets are hotbeds of asymmetric information. The party with the best information has the greatest power. Market making can provide that power.

All of this is well and good, says TED until the market makers’ balance sheets balloon into the trillions of dollars. At that point, market marking becomes indistinguishable from large-scale prop trading.

And in the CELF trade, for all Goldman’s protestations that it was making a market in otherwise-illiquid structured securities, in fact it was simply bidding on those securities, against other bidders:

Goldman’s actions in 2010 bear absolutely no resemblance to behaving like an agent when it purchased the outstanding CELF securities and liquidated them. It did not behave like a normal market maker, buying securities from one investor and selling them to another. It paid an arm’s length price, determined after an auction run by a third party, to the investor it originally sold the AAA rated tranche to. It then triggered the liquidation of the securitization by purchasing a majority stake in its equity. With respect to the seller of the AAA tranche, it acted as a pure trading counterparty. A principal.

I’ve since learned a few more interesting facts about this transaction, none of which make Goldman look particularly client-centric.

For one thing, the investor selling the AAA tranche was exactly the same as the investor from whom Goldman bought enough equity that it could gain control of the CLO and liquidate it. That’s really weird. It makes sense that a Dutch pension fund might want risk-free assets — pension funds are like that. But why on earth would they want a piece of the incredibly risky and volatile equity tranche as well?

Goldman’s trade, here, was to buy the AAA tranche at a discount, and then to also buy up the equity tranche — which, when added to the residual equity which Goldman already held, pushed its total shareholding over 50% and gave it control of the deal. It then found a couple of outside investors (hedge funds, I assume), to come along for the ride and to promise that they would vote as a bloc, and sold them a package of bonds and shares on the understanding that if the shareholders voted to liquidate, the bonds were sure to be paid off in full.

This is a really complicated way of adding value which could have gone much more easily, with much less use of Goldman’s balance sheet. Goldman’s financial advisers, with the best interests of their client at heart, could have looked at the pension fund’s assets and noticed that it had almost enough equity to liquidate the CLO and be paid off in full on its bonds. And looking into their own bottom drawer, they could have noticed that Goldman’s own shareholding would be enough to push the pension fund over the 50% mark.

The most obvious thing to do would be to simply tell their client that Goldman would vote with them to liquidate the deal, and thereby help them get 100 cents on the dollar for their AAA tranche. But since that might involve Goldman taking a loss on its equity tranche, the next most obvious thing to do would be for Goldman to sell its shares to the pension fund at a small premium, and simultaneously help the fund liquidate the CLO, making a substantial mark-to-market profit on the deal.

Both those actions would have been client-centric things to do, and neither would have constituted proprietary trading. But Goldman instead decided to wheel out more than a billion dollars of its own money to buy up the pension fund’s holdings at a discount and take the profit for itself. It was so much money, indeed, that Goldman felt the need to bring in outside partners to share the risk — risk it never even needed to take in the first place.

A clearer example of what TED is talking about could hardly be found. Goldman had asymmetric information: it knew that there was a good enough bid for the loans which made up the CLO that if the structure were unwound, all the debtholders would be paid off in full, and even the equity holders would make decent money. (I’m hearing that the equity holders walked away with 80 cents on the dollar.) Goldman then used its monster balance sheet to profit from that information itself, rather than simply sharing the information with its client and letting the client do the trade instead.

Goldman is adamant that what it did on this trade was client-focused market-making. “This is a good example of helping a client achieve its objective,” the bank told Jesse Eisinger, “and underscores the critical importance banks play in using their capital to facilitate transactions on behalf of clients.”

TED, on the other hand, sees things slightly differently, calling that statement “a patently disingenuous dodge” and “pure, unadulterated horse****”.

I’m tempted to side with TED on this one. Yes, the client’s objective was to get out of its position. But Goldman could have helped it do so more profitably than it did, and what’s more could have done so without using its capital at all. The only reason it didn’t was that it made more money this way round.


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


Interesting new facts, which support my gloss on the story. Goldman did not treat its counterparty as a client in any respect. Saying so is indeed vintage ordure de cheval.

Posted by EpicureanDeal | Report as abusive

You folks really don’t get it.

Goldman Sachs IS the client now, as is JPM and the other TBTF firms. The entities Formerly Known as Clients, as well as the taxpayer, now exist to serve the TBTF firms.

The entire global financial system is now structured to maximize the returns of the TBTF firms around the world, whether they be US, German or Japanese companies and to ensure that they are not injured even through their own monumental greed and stupidity. There is no other priority that takes precedence.

Posted by ErnieD | Report as abusive

Felix, you usually do a follow-up post/video in plain English for those of us who aren’t up to par with the finance lingo. Would it be too much to ask for you to do one for this particular story?

Posted by foodist | Report as abusive

Is it possible that the Dutch pension fund wasn’t actually allowed to take the loans on their books? I know lots of funds have very strict rules about the amount of capital they can allocate to lower rated securities. If that were the case it would explain Goldman’s insistence that the process was client driven. It might also explain why the fund had some of the equity. They wanted to make sure that the deal couldn’t be unwound on them putting them in an untenable position. Don’t know if that makes any sense or not, but it is an alternate explanation.

Posted by FreddieJJ | Report as abusive


Warn me so I can take a gravol! (sorry Felix, couldn’t help the little dig!)

Posted by hsvkitty | Report as abusive

“I’ve since learned a few more interesting facts about this transaction, none of which make Goldman look particularly client-centric.”

It should have been obvious to everyone that when Goldman got the government to take over AIG and then use AIG to bail out Goldman that Goldman is not very client-centric. By any measure, AIG was a huge client of Goldman, and for the short term gains of Goldman management, they threw AIG under the steam roller.

Goldman is not client-centric nor shareholder-centric. They are Goldman executive-centric.

And, by the way, they do not put their capital at risk. They risk the government’s capital.

Posted by OnTheTimes | Report as abusive

Felix, TED: good work, thanks

Posted by crocodilechuck | Report as abusive

Well lets do the sums shall we?

A1a notes got 6 month EURIBOR + 165bips, A2a got plus 150 and A2b got + 300. You imply that they took a 400 bips hit on the principal. In 2008 6m EURIBOR was averaged above 3% but lets assume 3 to make sums easy, so they will have earnt very roughly 37m on A1a note, 13m on A2a and 2m on A2b. In 2009, again lowball average is around 1.5% so they earn 25m, 8m and 1.5. 2010 again very roughly EURIBOR is 1%, so 21.5m, 7m and 1.3m and the haircut means a loss of 45m, total gain to fund is 71.3. Lets assume GS buys sub notes at par and sells at 80 plus gets that gain on the AAA tranche, GS makes 73m. So the difference to the client is around 2 million between your suggestion and what actually happened. In return for losing out on that extra 2m GS takes on a billion plus onto their balance sheet for just over a month whilst the underlying get sold/redeemed in the midst of a EUR crisis.

I can think of lots and lots of other reasons. Tax reasons for taking the loss on one bit and gains on the other. Regulatory restrictions of the amount of sub debt the client can buy vs AAA rated stuff. Not wanting to a billion plus risk in the middle of a EUR crisis. Certainly isn’t straight forward. Here is a crazy insane idea… why don’t you ASK the client?

PS here is the prospectus: Final-Prospectus

careful arithmetic was never my strong point so it is more than possible I made a horrible error, especially given the smug tone of the post…

Posted by Danny_Black | Report as abusive

foodist, it actually is not that complicated…

Basically what happens is that GS has a load of loans. Either loans it made or loans it bought. It doesn’t want to keep them on its balance sheet so it forms a new company. This company issues bonds with the promise to go out and buy these loans – it does not appear to be a static portfolio but the general characteristics are defined.

Thinking of the cash flow as water, there is a certain expected flow from the loans in terms of interest payments and then principal repayments. This is one bucket. There is typically some extra cash added in, this is the “over collaterisation” bucket, it is used to make up any shortfall in the loan bucket. These two then poor water into the third bucket. Imagine this bucket has holes in it. The first hole at the bottom of the bucket is the A1a note, water comes out if it first and the amount depends on the amount of water the loan bucket and O/C bucket are pouring in up to a limited amount. Above that hole is A2a and water comes out of there as soon as the A1a hole is flowing freely. Then A2b after. This is the AAA tranche. Any water left over is flows through the sub note hole. In principle, this structure should stay in place until 2020. It has to stay in place until Sept 2010 and then if people holding 50% or more of the sub notes vote so then the entire thing can be unwound – so the underlying loans are sold to someone else and the bonds are bought back with the money earnt from the sale plus the excess cash. The AAA tranche gets paid off in full – this is a requirement – and then what is left over goes to the sub note owners.

What GS did was buy back the AAA tranche and some of the equity and then use the right to unwind. Mr Salmon is suggesting that if GS was serving the client then they would have done it the other way round ie sold **their** equity to the client and left it to the client to unwind.

What makes this stuff complicated – and by the way gives it a major reason to exist – is tax and regulatory treatment. If the sub notes paid no interest and GS and others bought at 72.5 and sold at 80 then that is a capital gain whereas interest would be income and the tax treatment is usually different. The other issue can be regulation. The client might be the same but there might be different funds run by that client that invested, so maybe the ABC high-grade fund bought the AAA tranche and the ABC high-yield fund bought the sub notes. The high-grade fund won’t be allowed to buy the sub notes and the high-yield might have limits on their exposure to it. Also ABC might be the same company but the two funds might have different managers and will be separate legal entities.

Equity=subordinate bond, note=bond( normally a medium tenor bond ) in above.

Posted by Danny_Black | Report as abusive