Saturday, January 31, 2009

The ISDA CDS Settlement Auction is A Hidden Goldmine for Cash-Rich Accounts

When I discussed the basis trade opportunity, the conclusion was that arbitrage in the secondary (and by extension primary) market exists due to significant dislocations in liquidity. Sure, one can argue that immediate culprits for the arbitrage have to do with counterparty risk and funding costs, which makes sense, but those are merely derivatives of the liquidity disconnects between different brokers, their accounts, and any permutations thereof.

The problem with the basis trade, as Merrill and many others experienced, is that in the period of time before its par unwind, a lot of crazy stuff can happen that can force stop losses triggers, significant margin calls, and the overall liquidation of your business.

In the CDS realm a comparable "risk-free" strategy is participating in ISDA-mediated physically settled credit default swap auctions. These occur roughly a month after the ISDA council determines there has been a "credit event" in a given corporate name (most usually filing of bankruptcy, or some other default variations). As there will be many bankruptcies this year, and numerous physical settlement auctions, accounts that have liquidity can generate quick, practically riskless returns arising from a liquidity-fundamental value disconnect. In fact the most recent ISDA calendar indicates there are quite a few auctions coming up:
  • Millennium America, February 3
  • Lyondell, February 3
  • Equistar, February 3
  • Sanitec, February 5
  • British Vita, February 9
  • Nortel, February 10
  • Smurfit-Stone, TBD
One thing to note is that virtually any account which trades fixed income can participate in the auction, it is not limited to parties which have preexisting CDS relationships with the defaulted entity.

So in terms of a chronological summary (for the full chronolgy in the Lehman case, click here. Also props for whoever finds the rather big mistake ISDA made in compiling this schedule, hint: Tembec)
  • A few days before the auction ISDA prints the list of deliverable obligations that will be allowed to settle physically - many times these are most of the corporate bonds of an issuer, but in some very complex cases (Lehman and FNM/FRE most notably) only a certain percentage of bonds are selected.
  • The day before the auction, ISDA publishes the full list of participating bidders which will serve as the clearing agents to make sure all offers are exhausted.
  • On the day of the auction many things happen in rapid succession. The key items to keep track of are the submission of the amount of bonds one is willing to purchase and at what price, the determination of open interest (amount of bonds that have to be purchased in the auction by various dealers), and the final price, which Creditex posts at 2 pm on the auction date, which is also the close of the auction. It is important to understand that physical settlements are Dutch auctions: i.e. clearance occurs at the lowest price at which all net sell interest is absorbed. Therefore, a bidder does not stand to lose by placing a low ball bid as it may very well end up the critical one, and result in the final price.

Due to the significant residual overhang of protection sellers in the 2003-2007 period, most auctions end up with a net open sell interest. Let's take the case of Lehman:

After dealers submit preliminary inside bid indications, they are allowed to participate in the auction. The final bid price is based not on the inside bids but the actual limit orders (a combination of broker and customer orders) at which the Dutch auction clears.

The next important step is determining the Net Buy/Sell interest. As noted, most auctions are entered into with a net open sell interest due to excessive protection selling which is one of the main reason for the arbitrage. As presented, in the Lehman case, despite nearly $150 billion of defaulted unsecured debt, there was only $4.9 billion in net open interest (offered), proving all doomsayers of the CDS market wrong, as at its core it is a fundamentally very efficient market, regardless if it trades OTC or on an exchange.

After the net open interest is calculated, bidders submit limit bids which is where the liquidity arbitrage truly comes into play. The Lehman auction is probably not the best example of the arbitrage as two of the bidders had market-conflicting interests, although it will still demonstrate the point (Barclays which had legacy interests from its Lehman broker/dealer acquisition, and JPM which had the government's backstop at this point and likely other derivative interests in clearing bonds at a certain price). The chart below shows the cumulative total bid interest by pricing bucket (again for the full backup check out the Creditex link above).

Notable here is that the preponderance of bids are at ridiculously low levels, with the average of the total $132 billion notional in submitted bids (which is surprisingly close to the total amount of outstanding debt - perhaps dealers and accounts had assumed that as much as the total unsecured debt may be open to physical settlement) in the 2-3 cent range. Obviously many bidders were focused on the Dutch aspect of the auction, trying to aggressively lowball the market. As the final net sell interest was relatively low in comparison with the total bid interest, $4.9 Bn vs. $132 Bn, the clearing price was higher than the average. However, as mentioned above, JPM and Barclays, which likely had other motives in terms of procuring Lehman securities, alone accounted for $2.5 billion of the $4.9 billion in effective bids. Without these the final price would have had a 7 handle.

The key aspect of the ISDA auction, is that it significantly reprices the secondary market of the existing securities (which at least in theory is determined by fundamentals) based on an exercise in liquidity. It is feasible that with much larger open interest the ISDA auction may have cleared at 5, even 4. Why is this important? Because Lehmans' bonds in the days before the auction traded at markedly higher prices. As an example, let's take the 6.625% Notes due 1/2012, which in the three day prior to the auction closed with an average price of 12.5. (Chart below)

As representative Lehman bonds had been trading in low teens (in this example the 6.625% Notes closed at 13 on October 9, the day before the auction, according to TRACE) the highest bid bucket will almost certainly be below the market price. And this is where the arbitrage comes into play. Managers who wish to enter a specific bankrupt security don't have to buy it in the open market if there is an ISDA auction approaching - they merely have to submit a reasonably discounted bid in the ISDA auction in order to be able to purchase the security, by taking advantage of the liquidity dislocations in the market. Here, the ISDA clearing price of 8.625 was about 35% lower than the prior closing price (13) for an indicative Lehman issue which was a deliverable. This is also sometimes called the cheapest to deliver phenomenon as the clearing price will very likely always be lower than the market price of the cheapest trading security in the secondary market that is part of the deliverable schedule per ISDA.

The liquidity part of the equation comes into play when one considers that fewer and fewer dealers and funds have lots of spare cash lying around to participate in these kinds of auctions. Once an account submits a formal bid (bid and notional), it is binding. Which is why in many of the upcoming ISDA auctions (Lyondell will likely be the most anticipated one) a buysider may make off like a bandit merely by submitting a large, substantially out of the money big which gets triggered when the Dutch auction doesn't find filling bids on the way down.

Admittedly, there are many more nuances to the physical settlement process (could discuss offline time permitting), but in a nutshell it is a great opportunity for still liquid accounts to purchase bonds in bankrupt entities at a significant discount to market values. If one chooses so, these may be flipped immediately in the open market which still trades substantially higher than the auction results (in Lehman's case on the day after the auction, the 6.625% bonds closed at 9.6% allowing for a 10% return in one day).

There are odd exceptions: the Fannie/Freddie physical settlement auction ended up in some surprising results, where due to technical considerations, the subs ended up with a higher clearing price than the seniors (there is good literature out on the net as to why this happened). Additionally the final auction Creditex document provides some very good insight into which broker/dealers may have liquidity constraints. The Lehman auction specifically demonstrated that Morgan Stanley is severely strapped for cash: despite their inside bid/offer levels of 8.25/10.25, they indicated firm limit orders only below 5 cents. The upcoming auction results will provide a rare glimpse into just how cash limited (or not) broker dealers are currently: I would keep an eye out on Goldman, Morgan Stanley, Merrill and RBS. If the liquidity malaise is indeed spreading, funds will be able to take advantage of the brokers' troubles, and purchase securities at some very significant discounts to prevailing market rates.
Sphere: Related Content
Print this post


Broken said...

Good article on a market I have little exposure to.

Yu-Jia said...

Why do funds wait until the CDS auction to close out their long bond position? If it is well known that there are net sellers during the auction, the rational move for a manager is to sell out the long position before the auction. In this example, he can sell the bonds for 12.5, and then get paid out 100-8.625=91.375 on the CDS, for a net payout of 103.875.

Tyler Durden said...

good observation. i know a bunch who do just that. as to the arbitrage explicitly, it all depends on whether there is net buy or sell interest as you say. that is not known until all have disclosed positions. from that point on it "should" be market frenzy to take advantage of the forced selling by the initial allocators.

Anonymous said...

You say that due to the overhang of protection sellers, that most open interests are to sell. But in a CDS auction the protection sellers are buying the bonds and paying par. Not the other way round ???