Submitted by Peter Tchir of TF Capital Markets
CDS for Doomers, Politicians, and the Media
About the only thing that the doom and gloom crowd, the politicians, and the media all agree on is that credit derivatives are evil, unnecessary, ‘financial weapons of mass destruction’. With the European Sovereign Debt crisis escalating, the CDS market has once again become a topic of conversation. Many of the issues related to CDS that are discussed are old, misleading, or plain wrong. Here is my attempt to address some of the issues that come up most whenever CDS is mentioned:
- Credit Events
- Counterparty Risk
These are topics that need to be understood in order for investors to make informed decisions. I am not here to defend CDS as a product, but to try and shed light on the subject so that people don’t react to inaccuracies that cause them to make decisions based on incorrect information. Since so many journalists still feel that the investing public needs to see the boilerplate language ‘when yields go up, bonds prices, which move in the opposite direction, go down’ this may be an uphill struggle. But here is my attempt.
Credit Events Are Very Confusing
A Credit Event effectively terminates a CDS contract and locks in a gain for the Buyer of Protection and locks in a loss for the Seller of Protection. CDS contracts have value prior to a Credit Event and trade regularly and have a mark to market value. Investors who trade CDS generate P&L from being write or wrong in the direction of spreads on a particular contract. You do not need a Credit Event for a CDS to have value, but if there is no Credit Event during the life of the contract, then the CDS will expire worthless. If there is a Credit Event (as determined by the ISDA Credit Derivatives Determinations Committees) then the credit event settlement protocol is followed.
So that is why a Credit Event is important, but what is a Credit Event? It has nothing to do with the rating agencies. The rating agencies track defaults, but that is for their own purposes (to show annual default rates and historical ratings transitions, etc). While the rating agency determinations of defeualt are interesting, they have nothing to do with triggering a Credit Event for CDS. It is solely determined by the committee based on the 2003 ISDA Credit Derivatives and relevant Supplements. These can be found at ISDA Bookstore - ISDA Credit Derivatives Definitions, Supplements and Commentaries. The Committee is comprised of both market makers and investors in an effort to be fair and remove bias from their decisions.
The Committee has to follow the definitions to determine if a Credit Event has occurred. It is pretty straightforward for U.S. corporates. They trade “NR” or “No Restructuring” so the only Credit Events would be applicable are Bankruptcy and Failure to Pay. These are both pretty straightforward. Filing for chapter 7 or 11 would clearly trigger a Bankruptcy Credit Event. Failure to Pay is also pretty self explanatory. The company must fail to make a payment on an Obligation within the timeframe of any applicable grace period for that Obligation. The Obligation means Borrowed Money so is meant to cover money the company has borrowed, so as not to get triggered by a company failing to pay a contractor or service provider. There is a minimum size test so that the Obligation that caused the Credit Event is meaningful. There is also a requirement that the information is publicly available in order for a Credit Event to be officially triggered. In spite of some terms like Obligation, Borrowed Money, Publicly Available Information, etc, the Bankruptcy and Failure to Pay Credit Events are pretty straightforward and as far as I know, there has never been a serious dispute over one of these has events has occurred.
European Corporate CDS trades with “MMR” or “Modified Modified Restructuring” and Sovereign CDS trades with Restructuring and Repudiation/Moratorium as Credit Events. Europe retained a version of the Restructuring Credit Event largely at the insistence of its regulators. They were unwilling to provide regulatory capital relief for banks that bought protection on an NR basis. The ‘modified’ element means that if a Restructuring Credit Event occurs then the CDS contract has a limited number of choices of which Obligation to settle against. It attempts to prevent banks from ‘Restructuring’ a loan by extending the maturity while obtaining collateral and then delivering bonds that had been effectively subordinated on their CDS purchases creating windfall gains. It is definitely a bit tricky, but the limitations on what can be used in the settlement process of a European Corporate Restructuring Credit Event has limited any controversy.
Full Restructuring and Repudiation/Moratorium Credit Events apply to Sovereign CDS. Determining a whether a Restructuring Event has occurred still needs to follow the document, but as you can see in this analysis http://tfmarketadvisors.blogspot.com/2011/06/restructuring-credit-event-and.html it is less straightforward to determine whether the conditions were met or not. Particularly confusing is what was intended for voluntary restructuring. In any case, for all the noise about whether there will be a Credit Event on Restructuring for Greece, all you need to do is read the definitions included in the link and figure out how it will be interpreted. The intent of the ISDA contract will play a role. As much as some politicians and bloggers want to make the Credit Event determination a mystery, it is not a mystery. The outcome may be uncertain as it has to follow the legal document but that is no different than any other trade governed by a contract where sides may choose to dispute it.
It’s Impossible to tell where the Exposures are
The DTCC website provides some very useful overall market data. You can use this link http://www.dtcc.com/products/derivserv/data_table_i.php?tbid=5 to pull up the top 1000 reference entities. This is updated weekly.
This site shows what is the Gross Notional and Net Notional outstanding for the Reference Entities with the most CDS outstanding. It does not give the exposure of any individual institution to a specific name or to CDS as a whole, but knowing how much CDS is outstanding on a reference entity has a lot of practical uses.
For those not familiar with DTCC, it is the Depository Trust and Clearing Corporation. I believe back in the 90’s it became the primary clearing house for corporate bonds. Over time they started clearing credit derivatives. In order for their data to be meaningful, it must encompass the bulk of CDS trades, which I believe it does. In the 2003-2005 period, the CDS market was plagued by a lack of unsigned confirmations. The regulators were putting pressure on the industry to clean up the document backlog. There was real concern both inside and outside of banks that the product had grown too quickly and the systems had not caught up to the volumes. The fear was turned into action as the auto industry in particular had trouble. Volumes had spiked, unsigned confirms had grown, but people weren’t as focused on the risk until real fear of defaults gripped the market. Huge efforts were made to successfully clean up the document backlog and the DTCC played a role in ensuring that the problem of unsigned confirms wouldn’t occur again. The DTCC started settling CDS trades. The process had become more formal and it was in everyone’s interest to use the system as no one wanted the risk of unsigned confirmations. I believe that virtually all single name and index CDS transactions are booked via DTCC at this stage so their data captures virtually the entire market. I do not know of any significant players who don’t use the system, and cannot think of a reason they wouldn’t, so I am willing to rely on the data from DTCC as fairly all encompassing and would seriously question and CDS notionals outstanding that differ from those reported by DTCC as questionable.
So what do the terms mean? How can they be interpreted? This is probably easiest done by working through an example, and given the current news I can’t think of a better Reference Entity than the Hellenic Republic (aka Greece). As of June 10th the Gross Notional of Hellenic Republic CDS outstanding was $79 billion and the Net Notional was $5 billion.
The Net Notional is the amount of open exposure to a Credit Event on the Reference Entity. In the case of Greece, there is a total of $5 billion of open exposure. If a Credit Event occurs the ultimate transfer will be between investors who bought a total of $5 billion of CDS from investors who sold $5 billion of CDS. With a recovery rate of 50% that means there would be locked in losses of $2.5 billion for the sellers and locked in gains of $2.5 billion for the buyers of protection. In the grand scheme of how much Greek debt is outstanding, this is a manageable number.
Clearly the Gross Notional is much bigger than the Net Notional. There are several ways to increase the Gross Notional. If a market maker/dealer buys $10 million of CDS from one customer and sells $10 million of CDS to another customer, the Net amount reported would be $10 million (since that is the “naked” exposure) and the Gross would be $20 million as there are 2 trades outstanding of $10 million each. If all trades were so simple you would expect the Gross to Net ratio to be 2:1. Not all trades are that simple. At least as common would be for Customer A to buy CDS from Dealer X, and Dealer X to buy from Dealer Y who then buys from Customer B. You still have a Net of $10 as only customer A and B have naked postions, but now the Gross is $30 since there are 3 trades in the system.
When a name is active, you will find that the street trades a lot with each other. This interdealer trading increases the Gross amount outstanding as dealers don’t do assignments or unwinds of old trades on a daily basis. All else being equal, I would expect the ratio of Gross to Net to increase for active names since the street will have high volumes trading amongst themselves while managing risk from providing liquidity to clients. Gross Notionals like this should not scare investors. The dealers will have collateral provisions amongst themselves under their ISDA Master Agreements so it would be surprising if any one dealer had too much exposure to another dealer on a single name basis or across the product. Furthermore, the dealers periodically work on collapsing their trades. They periodically submit trades to something like trioptima and the clearing corporations to “collapse” trades. If Dealer A bought from Dealer B who bought from Dealer C, after running a “collapse” methodology, you would only have 1 trade where Dealer A buys from Dealer C. The size of the institutions and the creditworthiness of them makes this sort of process work effectively. At times of crisis, say Lehman for example, they would not be included in a way that would force any member to take exposure to them.
As a Reference Entity becomes distressed, such as Greece, the dealers run these netting/optimization programs more frequently to reduce the number of trades outstanding.
Isn’t it possible that a hedge fund or some entity stands in the middle of a lot of trades? If this was the case, I would be more concerned about Gross Notionals being large, but I don’t think it is the case. Funds have to provide collateral when they do CDS trades. The amount varies on the Reference Entity and the fund itself, and the ISDA agreements it has in place with dealers. These collateral provisions are generally much larger when you have sold protection, though some counterparties are required to post collateral when they have bought protection. At first that might seem strange, but someone who is paying 1400 bps for 5 year Greek CDS could lose a lot of money if it gapped back to even 700 bps. So a fund would have to post collateral on every trade they wrote protection on, and possibly on trades they bought protection on. It is not in their interest to tie up so much collateral. It is also not in their interest to take on exposure to multiple counterparties for no good reason other than to have a lot of trades on. So customers typically do an ‘unwind’ or an ‘assignment’ when they trade out of a position. If they bought from Dealer A, and then sell back to Dealer A, they would “unwind” the old trade. There would be no trade in existence anymore which makes sense and is ideal for the entire system as there is no residual counterparty exposure – just like if the client had bought and sold a bond. If the customer sold to Dealer A and is now buying from Dealer B they would initiate a request for an “Assignment”. Assuming Dealer A is willing to take Dealer B’s credit risk, and vice versa, the client would “step out” of the transaction and Dealer A would face Dealer B. That trade would then be put into any future dealer to dealer optimization/netting program. The process has a high degree of automization and is in everyone’s interest to use since it mitigates counterparty exposure outside of the dealer community.
There is another way to get relatively large Gross notionals compared to Net notionals, and that is when “curve” trades are active on a name. If a client buys 1 year CDS and sells 5 year CDS they have put on a “curve” trade. If there is a Credit Event within the 1 year time frame they have no net exposure to the name. A trade like this would create $20 or $30 of Gross notional depending whether both trades are done with the same dealer but $0 of Net exposure. The client frequently does both “legs” of the trade with the same dealer because the dealer can charge less collateral so long as the trades are both on. The client will have some risk based on the price of 1 year CDS versus 5 year CDS, but this is less than the risk of being outright long or short. If there is a Credit Event and the client has both trades on with 1 dealer, they should collapse at equal value assuming they were done under SNAC protocol which almost all CDS trades have used since it was implemented in April 2009.
So the real exposure to the system is Net Notional as that represents real open positions and Gross Notional can be interesting but should not be viewed as fearfully as it sometimes is as most will be netted at time of Credit Event.
There is hidden Daisy Chain Counterparty Credit Exposure
There still seems to be so much talk about how a Credit Event on one name could have a ripple effect causing counterparties to fail on their obligations. I think this risk is severely overblown.
The risk of some daisy chain of failure is tied to the Gross and Net Notionals above. In an ideal and simple world (think exchanges) the Gross Notional would equal the Net Notional. There would be no one who is merely standing in between trades (except the exchange). That is not practical as not all clients deal with all dealers, etc, but I think we can quickly show that the risk of a daisy chain collapse is fairly remote.
Lets start with the customers as they are the easiest to analyze. They will either be short, long, or have some form of curve trade. They are highly unlikely, due to collateral requirements, to be flat the name with lots of trades on.
If the client has a curve trade on with one dealer, it should collapse as post a Credit Event all points on the curve should trade to estimated recovery and then to ultimate recovery. The ISDA Master will ensure this netting occurs so it won’t be the case where cash transfers on one leg but not the other, so no counterparty risk arises from this situation.
If the client has the legs on with two different dealers it is no different than the client being long with one dealer and short with the other dealer. Those dealers should have appropriate collateral in either case. It is up to dealers to charge appropriate collateral to clients for the positions they have one with them. In spite of the hype the dealers have been pretty good at managing their counterparty exposure. Not only would they have to be under collateralized, but the counterparty would have to post any further money required at time of settlement. Maybe it is hard for people to believe, but the banks have done a good job of managing exposure to hedge funds. The only 2 big examples I can think of were LTCM and AIG. LTCM shocked the markets partly because the banks didn’t realize how much of the same business they did with every dealer and they had negotiated some sweetheart terms in their ISDA’s. Credit departments became even tougher on hedge funds after LTCM, and in spite of no widespread losses from hedge fund counterparties during the 2008 collapse of credit markets, they have made the terms even more difficult in most cases. The other problem was AIG FP. There banks gave too much credit to the high rating and failed to estimate how much collateral would be posted at the worst possible time. I’m still shocked banks got bailed out by that decision, but my understanding is they are far more careful with even AAA counterparties than they were before.
I may have more faith than most people that the banks are managing their counterparty exposure to end users well, but my experience is that they have, and there is relatively little evidence pointing to banks having big losses due to mismanaging their end user counterparty credit exposure.
What about the interdealer exposure? Again, they will net that down through optimization/netting programs. Right now none of the dealers appear to be in trouble and since most will have small Net positions it is just a matter of money flowing through the chain. Could someone be sitting on 20 billion of Greece gross exposure? Sure, but if JPM is, that means they expect the correct amount of money to come in and to pay out the correct amount. I have addressed why I believe they will collect what is owed from their end user customers and there is no reason to believe they won’t collect what they are owed from the interdealer community.
The daisy chain failure of counterparty after counterparty sounds exciting, but I think that risk is well managed, and to the extent one of the dealers gets in trouble again, the remaining dealers will work around it. The Fed and every banking analyst was horrified about the prospect of Lehman, as such a large CDS counterparty failing. In the end, dealers seemed to manage that risk fairly well. There are few if any stories of banks being devasted by losses resulting from Lehman going away as a CDS counterparty (far more stories of big losses due to owning too many Lehman bonds), and the lawsuits revolving around Lehman’s bankruptcy seem to focus on such “simple” things as repo and which entity Lehman had moved money to/from. I am not aware of any big CDS lawsuits outstanding. CDS may sound big and scary but the reality is the losses that hit the market were from traditional problems such as being long bonds and being wrong or having cash tied up in an entity that defaulted where the terms were one sided. I am not discounting the issue completely but this is should not be the focus of investors trying to analyze the impact of Credit Events.
There is no price transparency
This is a bit tricky. For the average investor there is limited price transparency, but for the people in the market there is a high degree of price transparency.
The CDX Indices are quite liquid. They are quoted on live market in big size all day long by multiple dealers. At any given time an investor could buy or sell $250 million of IG16 on a ½ bp market and trade $50 million of HY16 on an 1/8 th of a point market. The liquidity in quiet times is even higher. That is far more liquid than any other part of the credit markets. It is highly transparent to the people involved in the credit markets but can be found on the Mark-it Website. The indices are posted on their home page, but this link can take you to pricing for the 5 year point of most Reference Entities. http://www.markit.com/cds/most_liquid/markit_liquid.shtml
On the single name CDS, the 5 year point is the most useful and accurate. Generally the more active a name is, and the bigger the market cap of its debt/equity, the more accurate the pricing is. Just like many high yield bonds, but even some investment grade bonds, are “quote only”, not all CDS quotes are as real. You can be pretty sure that the 5 year CDS spread on the markit site is close to levels that a CDS trader could execute on. For more off the run names and smaller high yield names, the price might be more indicative than real (an actual bid or offer could be a bit away or you may have to work an order to execute at those levels rather than demanding the dealer provide liquidity). Although that sounds scary, the Gross and Net Notionals on those will be small so any impact is minimal and the participants in the market manage around it quite well.
As you move away from the 5 year point, I am not sure where you can get publicly available information on prices. I use Bloomberg so I haven’t spent time looking, so it may be available and I just haven’t found them.
Shorter dated CDS tends to be less liquid than 5 year and the quoted prices can get wide, but that reflects the binary nature of stressed names more than anything and isn’t dissimilar to the cash bond markets where the liquidity in short dated paper often dries up in times of stress – with memorable exceptions being the day before Bear and the day before Lehman where volumes sky rocketed.
The sort of trades AIG was doing were not transparent, but they have little to do with anything that I have talked about here – single name corporate and sovereign CDS and the associated indices.
Transparency could be better, it might even help outside analysts, but at the same time, the transparency for people in the market is quite good, particularly in the indices and at the 5 year point on the curve.
Some effort could be made to make the market more transparent, but the same could be said for the corporate bond market. Corporate bonds themselves are not particularly liquid nor transparent. Prices on corporate bonds, even with TRACE aren’t that transparent. All you have to do is look at small lots on TRACE and see the individuals still pay a massive bid/offer on corporate bonds, and if a bond doesn’t TRACE it is hard to find the price. Some market professionals have argued that TRACE has done little for transparency since individuals don’t seem to notice and professionals already know where bonds are quoted, so all it has really done is make it more difficult to move large positions.