Settlement Auction for Lehman CDS: Surprises Behind

The surprise: no price support at pre-auction bond prices. (No inside bid exceeded 10%, and the inside market midpoint was 9.75%.) The relief: despite a substantial net open interest to sell bonds for physical settlement, barely half of which would have cleared at the inside market midpoint price, the excess supply was soaked up by limit order bids within the inside market's spread, resulting in a final price of 8.625%. There was ample additional demand at slightly lower prices, implying that at least some dealers' bids during the inside market phase really reflected a willingness to purchase plenty of bonds at their bid price.

(See previous article: "Settlement Auction for Lehman CDS: Surprises Ahead?")

A closer look at the limit order phase shows us who is starving for liquidity and who thinks there is money to be made by buying and holding these bonds. Here are the net trades resulting from the auction, ordered from most bonds sold to most bought. (Keep in mind that many of these bids and offers were placed by dealers on behalf of their customers, so they don't entirely reflect the trading strategies of the dealers themselves.) All volumes are notional. Net sellers first:

Deutsche Bank sold $870m. They didn't place a single limit bid above 8.5, and would have been net sellers at any price above 6. They would have been happy to take plenty at 4 or 5, though.

Credit Suisse sold $705m. This is net of $50m for which they (or more likely one of their customers) were willing to pay the top price possible (10.75) in the limit order phase. Otherwise they didn't have any interest above 8.5, and would also have been net sellers at any price above 6 (at which they were willing to soak up over a billion dollars' worth!).

Morgan Stanley sold $430m, similarly net of $50m of bids at 10.75. They wouldn't have been net buyers at any price above the absolute minimum possible bid (0.125), suggesting that their bid of 8 during the inside market phase was more or less a sham. The interpretation that comes to mind - about which I have no inside information whatsoever, mind you - is that they simply don't have the liquidity to play the "investment bank" game any longer.

Goldman Sachs sold $376m or so; they got there by offering to sell $1470m and then buying up over $1 billion of that ($150m effectively bid at "any price from the 10.75 maximum down", $675m via limit orders at the price set by inside market fixing, and the rest farther down the limit bid sequence). This seems to reflect their huge role as a dealer rather than any great appetite on their own behalf; they would have been net buyers at 8, but the bid that looks to me like Goldman's "we'll take it if no one else wants it" backstop is at 1.25. That isn't the derisory bid that Morgan put in, but it doesn't reflect a lot of willingness to scoop up a chunk of the ultimate Lehman liquidation either.

BNP Paribas sold $370m; their inside market bid of 9 wasn't backed up by any real participation in the limit order phase. They or a customer would have gone long at 2.125, and taken some more at 0.25, and that's it. They mostly just wanted to unload the stuff.

HSBC sold $157m (net of $30m bid at the inside market fixing). Otherwise they didn't bid at all. Some market maker - especially since they were the highest bidder in the inside market phase.

UBS sold about $83m; they would have been a much bigger seller at any price above 9, and would have been neutral at 8.5, substantial buyers at 8, and massive buyers at 6.

Banc of America sold just under $78m. They were clearly in no position to backstop the market but did bid for another $15m at each price point all the way down to 4 (resulting in a crossover to being a net buyer at 7.875). They don't have a lot of cash to play with, but at least they're playing like they believe in the game.

RBS sold $31m (after $160m of buy-back, probably mostly on a customer's behalf); despite having bid on the high side at 9.25 during the inside market phase, they wouldn't have taken any more back at any price. Another sham bid, if you ask me; but no surprise given that they seem to be on the verge of semi-nationalization.

Citigroup sold $24m, net of a couple of small purchases and a big $500m bid (possibly on their own behalf) just below their inside market bid. They would have taken another $500m at 8.375 and kept on buying all the way down (with a real monster bid at 6.625). Who knows where they think they would get the cash; maybe someone is bidding through them on a disproportionate scale.

And now the net buyers:

Dresdner bought $75m, probably on others' behalf; although they were one of only three bidders offering to buy during the initial physical settlement, the quantity was small at $30m and the bid was low at 8. However, they were willing to soak up supply if others were to have dropped out, starting just below that bid of 8 and continuing down to their big backstop at 1.375.

Merrill Lynch bought $529m; they were the only dealer to end up on the opposite side from where they started (having offered $141m in the initial physical settlement), thanks to a $670m bid at 10.25. (Presumably that's not their own trade, as they were among the bottom bidders at 8 during the inside market phase!) They would have bought more at 8.5 or below, but didn't even bother to put in a shoot-the-moon bid near zero.

Barclays bought $1210m, most of it with bids at the the price set by inside market fixing. They (or their customers) would have taken some more at lower prices, but their bid volumes peter out on the way down; they don't seem to have been interested in mopping up a potential excess.

That leaves JPMorgan Chase, the biggest buyer at over $1310m. They were by far the biggest bidder ($612m) in the initial physical settlement, and just kept on soaking it up; their bid for $500m at 8.625 wound up setting the final price. They were the one net buyer who would have kept on buying and buying and buying at prices not much lower than this - although there was plenty of other demand in this price bracket and, as previously mentioned, UBS would have started stepping in for real at 8. Doubtless there were lots of outside bidders using JPMorgan as their dealer, but evidently Chase trades heavily enough that they could have made this whole market themselves.

In short, this auction was basically successful at clearing the market for physical settlement at a credible price (contrary to outsider expectations though it might have been). However, without Barclays (BCS) (the purchaser of Lehman's assets) and JPMorgan Chase (JPM) (the Treasury's agent in winding down Lehman's trades), it would have been a very different story. (I wonder what combination of them has inherited trusteeship for all of the Lehman-created structured vehicles that issued CDS on Lehman itself?) The auction also seems to have provided a rare view into the actual trading behavior of the top broker/dealers, including apparent confirmation that Morgan (MS) and Goldman (GS) (not to mention RBS (RBS)) are seriously starving for cash.

Disclosure: none

Comments

[Amicus:]

Michael, great write-up. Thanks!

After looking over the Freddie/Fannie anomaly and reading through this, I have an alternative conjecture for your thesis about the "split result" in the last auction, in which the stop in one part of that auction trailed out at 99-something.

Could it be that the derivatives people participating in these auctions are simply ... not as skilled and used to running them as are, say, the primary dealers' desk?

While not inconsistent, per se, they look ... incomplete, in a technical sense of the term. I mean, look at Barclay's bids. Under what theory would bids peter out with *declining* price? Ditto Merrill Lynch. Compare that the Bank of America - their desk are willing to take the same amount of bonds at almost any price - with a bidding strategy like that, one imagines that ISDA should create a sidecar category for "non-compete" bids...

To be charitable, maybe it has to do with the structure of the auctions or the structure of the markets (can't treasury dealers go short in advance of auctions, in order to protect themselves?).

Last, keying off your insight about the impact that the net open interest might have on auction results, can we suggest that the reason that the reason the mid-market average came in below the pre-auction trading prices is that the net open interest was to sell bonds? Just under 60% of the satisfied limit orders were from three firms..

Last, I don't understand the Barclay's limit bids, in relation to the size of their physical settlement request. They put in a physical settlement request of $130m to buy, with a dealer bid at 8, then "won" $1,080m - almost 10x - with a weighted average bid of 9.47 in the limit order.

2008 Oct 12 11:43 AM

[whidbey:]

The perception that this was a rational trade session is probably false. The whole process was likely a mismatch of capacity v. value perceptions. This gives little confidence that CDS are understood by the public or the broker/dealers who have very little notion of how to value them. All to our peril.

2008 Oct 12 12:49 PM

[Amicus:]

It is interesting, to note, also, that the bid-offer spread in the first part of the auction went way up, from the Fannie/Feddire to the Lehman auctions, from an average of circa 2% to one circa 20%.

One wonders if this isn't partly related to the $640,000 quasi-penalty Goldman paid in "adjustments" in the last round (the inside market quote size also was cut in half). If so, this might explain Barclay's bid at 8 that I wondered about aloud above.

Whidbey, I'm certainly not an expert, as some of my questions probably show.

At this time, it's not clear to me how or why the valuation of CDS would be affected by the success or failure of these auctions - don't we theoretically like the price of credit to the probability of default, not to potential recovery rates? Obviously, I need to hit the books.

As for the auction itself, I've been through MarkIt's Auction Primer, to the point that I think I understand the calculations, but not the rationals (except at the big picture level of trying to set a settlement price).

For instance, when I first looked it through, I thought that the approach for any one firm would be to try to "win" in the limit phase about the same face value that was indicated to create the 'open interest' in via the first phase.

In that vein, I'm not sure, right now, what it "means" to offer $141m, but then end up with "winning" bids for $670m. Client orders? Why wouldn't they be part of the initial request, so that the total 'open interest' was larger?

I'm sure there is a logical explanation and it's just too late a night for me to figure it out. G'nite and g'luck.

2008 Oct 13 12:03 AM

[Amicus:]

Also, DTC has released this important note about settlement:

www.dtcc.com/news/pres...

2008 Oct 13 12:23 AM

[tragicomix:]

This is amazing !!!!

Bo Peng says that: "On Oct 21, somebody [group A] will have to pay somebody else [group B] billions in cash to settle Credit Default Swaps (CDS) on Lehman. Estimates on what this entails range from $100 billion to as much as $400 billion.".

Felix Salmon has $6 billion net pay out.

Michael Edwards is very satisfied that it all worked out right but does not say what the net pay out is.

The head of the German Financial Regulators said in a Reuters story today that "We're still licking the wounds of Lehman," said Jochen Sanio, president of the German Federal Financial Supervisory Authority, at an international banking conference. "It caused international damage of $300 billion outside the U.S."

Is the panic injection of equity by OECD member countries is based on the hard reality of a $300-400 bn cash payout (according to Bo Peng) or may be a misunderstanding (possible inference from Felix Salmon) ?

One thing is clear. Don't let rocket scientists anywhere close to where their bets co-mingle with normal fund flow ! Merriwether, Salomon, LTCM and now everybody has made the grievous error !

2008 Oct 13 02:28 PM

[AnonymousAlso:]

Maybe Morgan & Goldman as part of this deal to become BHC are minimizing their particpiation in this area? Goldman just got $10 billion + and one would guess Buffet who appears to dislike these kind of things would have looked carefully.

In addtion the centeral warehouse / clearinghousefor CDS comments that the net transfers on Lehman will only total about $6 billion. Does that jib with what you see?

2008 Oct 14 01:06 PM