Lessons Learned

A week has passed since the auction for Greek CDS. Perhaps it’s now time to reflect on the credit event process. Toward that end we wanted to share our thoughts in a combined derivatiViews and media.comment post, and we also encourage our readers to offer their views.

In our minds, the most striking thing about the entire situation was the wholesale shift in sentiment regarding the potential risks of a credit event. In the space of a few months, it went from being a big issue to a non-issue (though it really should not have been an issue at all). We – and anyone who looked at the DTCC’s trade repository website – knew all along that the level of Greek CDS exposure was relatively small. In addition, while it was published in aggregate on the DTCC’s site, it was known on an individual firm level to regulators. The credit event truly was a non-event.

One of the major reasons why it was a non-event is because of the significant amount of work that ISDA and the Determinations Committees have put into ensuring that the credit event process is fair and robust. This process has been tested many times since it was introduced a few years ago and continues to work well for all market participants.

Our biggest disappointment throughout the process was the lack of understanding by some of two important points about the credit event process. The first point relates to the structure, composition and workings of the Determinations Committees. Apparently, the fact that the names of the individual firm representatives serving on the DCs are not disclosed makes them “secretive” to some. This is despite the fact that the names of the firms serving on the DCs are public, their votes are public, and the rules governing how the DCs function are public. It’s important to note that the individual firm representatives can and do change from credit event to credit event; there is no “list” per se.

The second point relates to the nature and definition of a credit event. As we said repeatedly, particularly here, a contract is a contract. One can speculate about what might be or what should be – and many did. But we repeatedly urged people to read and understand the contract as written. If they had, then there would have been little surprise that the DC could really not act until the collective action clauses (CACs) were invoked by the Greek government. This important step meant that the Greek restructuring was binding on ALL holders, which is a condition required for a credit event to occur under the restructuring clause. In addition, until the Greek government acted – and posted their action in the official government gazette – the CACs were not officially invoked. This too is required before a credit event can be declared. That’s because the DCs do not vote prospectively on credit events.

The Greek credit event also demonstrates to ISDA that we have more work to do. Some market participants legitimately raised the question of whether the package of obligations issued in exchange for old Greek bonds should be considered in the Greek credit event auction, arguing that this was the “right” economic result. Yet among those obligations were certificates issued by the European Financial Stability Facility, not the Greek government, so the package was not considered in the auction.

The fact that the package was not included in the auction was picked up in the blogosphere as evidence that CDS are somehow fundamentally flawed. We beg to differ with that broad characterization.

We believe that it is important to adhere to the terms of contracts as written and agreed between parties as to do otherwise would adversely impact the market. Also, we knew there would be good deliverables for the auction. But we at ISDA also have a long track record of learning from and adapting to market experiences, particularly ones as significant as this.

We are also committed to considering changes going forward, not just for new contracts, but where there is market consensus for a change, for existing contracts as well. One need only look at the 2009 Big Bang Protocol for evidence, when the structure for CDS for both new and existing CDS was agreed broadly by market participants.

Whether, when and how to change the contract to address this recent experience is already being debated by market participants. As we have on many occasions before – for CDS and for the whole range of OTC derivatives – ISDA will play a central role in facilitating the evolution of products that we believe are an essential part of the fabric of the credit markets and of the financial system as a whole.

Stay tuned!

When It Comes to Sovereign CDS, Collateral is King

We have talked a lot about sovereign CDS exposures being largely collateralized. What exactly do we mean by that? And why is it so important to understand how collateral works?

Exposures between two counterparties under an ISDA Master Agreement are typically subject to one of ISDA’s credit support annexes. Our margin survey indicates that over 70% of derivatives exposure is subject to these arrangements. But some of the entities that are users of other types of derivatives ‒ sovereigns, supranationals and corporates ‒ are not active in CDS. As a result, well over 90% of CDS are subject to collateral arrangements, and these arrangements are virtually all two-way (i.e., either party could post collateral to the other based on the mark-to-market value of trades between them).

With the standardization of CDS contracts resulting from the 2009 Big Bang Protocol, there are now two standard coupons on CDS, 100 basis points for investment grade credits and 500 basis points for high yield names. An upfront payment is made by one party to the other to reflect the present value of the difference between the market rate for buying protection and the standard coupon. Where the reference entity is distressed, a significant amount is paid by the buyer of protection upfront because the market rate for buying protection greatly exceeds the coupon.

This upfront payment feature of the CDS market is distinct from other derivatives markets. But once the upfront payment is made, the collateral practice is the same. The trade is marked to market and collateral is posted. For a distressed reference entity, the protection seller would have to post collateral, but the amount of that collateral would, at least initially, be more or less equivalent to the upfront payment. In effect, the upfront payment from the buyer to the seller becomes the collateral that the seller posts with the buyer. Subsequent, incremental fluctuations in market value will lead to more collateral being posted or some collateral being returned.

To understand the implications of collateral arrangements, let’s take a simple example involving Greek CDS. Let’s assume that Bank A is a net seller of five-year protection and that all of its Greek CDS trades are with one other bank, Bank B.  In this case, Bank A would have posted collateral, primarily cash, to Bank B in an amount equal to the mark-to-market value of the CDS trades. 

As the likelihood of default increases, the value of the contract will increase as well and more collateral will need to be posted, a process that happens daily. So based upon current prices for five-year Greek CDS, Bank A will be posting 62% or so of notional against the Greek protection it has sold. Each day, assuming conditions get worse, the amount of collateral increases. The risk for Bank B is that either Bank A defaults on one of the daily incremental collateral calls or a Greek default occurs and the price of the CDS increases greatly. However, this latter event would hardly constitute an extreme “jump to default” situation, such as where an otherwise creditworthy entity defaults out of the blue, generating a collateral call of 50% or more. So Bank A’s daily collateral requirement will be relatively modest and the replacement cost to Bank B will be modest if, indeed, Bank A should default for whatever reason.

Counterparty risk management for CDS involves assessing the exposure that could arise between one counterparty and another due to a jump to default by a reference entity or a counterparty default on the CDS. How much does it cost to replace the defaulted positions as well as the shortfall in collateral? There is no unsecured exposure before the default. In Bank A’s case, it already has posted 62% against its sold Greek protection.

Collateral management between two dealers is more complex, of course, because all their OTC derivatives contracts are covered by one collateral arrangement. But the principle is the same: how much collateral was not delivered and how much will it cost to replace the positions? Collateral arrangements between dealers are also very efficient due to netting arrangements that enable in the money positions to offset out of the money positions across asset classes.

One final note: Consider what happens when, as is far more likely than not, Bank A performs its obligations under the Greek CDS. When Bank A performs its obligations on the Greek CDS the balance on its collateral arrangements with Bank B will be altered. After payment on the Greek CDS, the net mark-to-market position shifts. Because the Greek CDS trades are settled, Bank A will be able to demand collateral to be returned to it by Bank B.

Collateral is an incredibly powerful and dynamic tool and understanding how it works should, we believe, provide a great deal of comfort to both experienced and casual observers of the sovereign CDS product.

The New York Fed Report, Hedging and Market-Making

Last week, derivatiViews commented on the New York Fed’s study on the CDS market. The study confirmed in many ways our own research and provided some important new data on market participants and their activity in the CDS market. We only wish we could see more OTC derivatives analyses from the New York Fed and other, well-regarded regulatory bodies. This week we address market-making and hedging, another topic covered in the report.

In Section X on Hedging in the New York Fed report, it was noted that when dealers executed large single-name CDS trades for customers, they did not engage in any offsetting trades on the same day 47% of the time. However, this was an average number and needed to be examined further. The report showed that for actively traded reference entities, same day hedging occurred 79% of the time. As might be expected for less liquid reference entities, there was less same day hedging – 55% of the time for less frequently traded entities and only 44% of the time for infrequently traded entities. And remember, there are only 10 trades a day in actively traded entities globally, four in less frequently traded entities and less than one trade per day in the infrequent entities.

Following this analysis of the market was a single statement in the Fed report:  “Our analysis seems to suggest that requiring same day reporting of CDS trading may not significantly disrupt same day hedging activity, since little such activity occurs in the same instrument.”

Some, though not all, commentators pounced on the statement, concluding same day reporting for large trades was the way to go. We were surprised at this. Indeed, we are still surprised. But one of ISDA’s responsibilities is education so we will educate.

For one, commentators may not have read the next sentence. It urged policymakers to gauge what impact greater post-trading transparency would have on dealers as dealers’ current hedging approach facilitates trading in the CDS market. Nor might they have understood the statement in the Executive Summary: “The low trade frequency of most CDS… combined with relatively large trade sizes highlights the important liquidity providing role of the market maker…”

So we thought we’d go back to square one and talk about market-making. We’ll start with a government bond desk. Say a client asks for a bid for an off-the-run 15-year Treasury bond. A bid is given and the dealer buys the bond. Does the dealer immediately try to sell the bond? Probably not. Suppose, just as it has purchased the 15-year bonds, it sells a 10-year Treasury note to another client and has thereby hedged its exposure to interest rate movements. The dealer may carry both the long and short positions on its balance sheet and will only unload the positions if it is concerned about curve risk, balance sheet usage or financing availability. Naturally, the dealer’s best outcome is to sell the 15-year bonds to one client and buy the 10-year notes back from another. But the willingness to hold positions explains why dealers carry billions of dollars of government bond inventory.

A similar process occurs for corporate bonds and CDS. Dealers have single name, industry and rating limits for credit product as well as aged inventory policies. Dealers are meant to facilitate client business. They have merchandise for sale – inventory purchased from clients – and add to that inventory when clients trade with them. It is no secret that, after providing liquidity to a client, a dealer may require a prolonged period to work out of the position.

Let’s look at an example of how this applies to a large CDS trade. Assume a dealer sells $20 million of protection on an infrequently traded reference entity, one of the 1,300 entities which the Fed has calculated trades less than once a day. What does it do? If the dealer immediately shows an interest in buying protection to brokers, other dealers will know it has a position. With less than one trade executed per day globally, the dealer needs to keep its position quiet. It may show a slight interest in one way or another to hedge part of the position but it must be careful or the market will trade against it. It may “hedge” its position through an index trade but it will try to work out of the position over time unless, of course, it does not like the credit. Then, it buys protection at whatever price makes sense to it.

What does this say about same-day reporting for large single-name CDS? It does not mean the full size and price details should be revealed immediately or even with a delay. Perhaps a good start is the TRACE system. TRACE requires disclosure of corporate bond trades within 15 minutes of execution but trade size need not be disclosed if it exceeds $1 million. The corporate bond market is more active although it trades in smaller sizes per trade. It respects the role of market-making and has protected it to ensure clients can buy and sell securities at reasonable prices. TRACE is not perfect but we hope reporting regulations for CDS markets will provide comparable protection of market-making.

We hope our little class has been useful and thank the New York Fed for their analysis. We know of more than one regulator that still owes the marketplace a cost-benefit analysis or two.

CDS Unveiled: The NY Fed’s CDS Report

The Federal Reserve Bank of New York issued a paper on the CDS market this week that was quite thorough and professional. We found it interesting on several fronts – so much so that we’ll devote two derivatiViews posts to it. Today’s will focus on market size and we’ll agree with most of what the report reveals. We’ll try to show how market-making works and how dealers offset risk as well as the importance of reporting delays or incomplete trade reporting for large trades. (Readers should also know that our press colleagues have discussed briefly the media reaction to the report in our media.comment blog.)

The Fed paper examines three aspects of CDS market size: the frequency of transactions, the average size of transactions and finally the number of participants. We will focus on the single-name CDS market as that has clear analogies in the corporate bond market and it will keep our story short and sweet.

Regarding frequency of transactions, the report confirms what ISDA and other commentators have often said in the past. Single-name CDS do not trade very often. In fact, the report shows a total of 3,000 single-name trades a day – both corporate and sovereigns. This global volume compares with volumes in the US corporate bond market alone (which excludes the large Eurobond market) that are five or more times as large. Over a 12-hour trading day, single-name CDS trade only 250 per hour. Apparently, CDS traders do not do very much trading.

But that’s not all. The report divides the 1,554 reference entities that traded during the three-month study period into three categories: the top 48 “actively traded” reference entities; the next 219 “less actively traded entities; and the last 1,287 “infrequently traded” entities. How much do each of these entities trade each day? The answer is: not much. The actively traded names trade only 10 times on average. Less actively traded entities trade only four times per day while the largest category by far, the infrequently traded entities, trade on average less than one time per day. Each of these figures confirms another fact about the CDS market. Not only is the market small in the aggregate but it is small with respect to every single name as well.

The report also contains useful information about transaction size. In the single-name market, the average size trade for corporate CDS denominated in US dollars is $6.7 million and for euro-denominated CDS it’s €5.9 million. Sovereign average size is larger – $16.7 million and €12.5 million. Five percent of US dollar corporate trades are $20 million or higher while the same figure for sovereign CDS is $50 million and €50 million. We do not have comparable figures for the US corporate bond market but knowledgeable participants have assured us $20 million trades do not make up 5% of the market. These figures point out what other studies have shown. The CDS market is a wholesale market designed for large players.

Finally, the last characteristic of market size is the number of participants. Here we disagree with some of the comments in the report, but are grateful for the data the NY Fed compiled. The report finds a “broad level of participation in the CDS markets” and cites as evidence the fact that there are 50 to 100 unique market participants trading daily in single-name CDS. We recognize the report uses the word “broad” in a number of ways, but how can 50 to 100 participants be categorized as broad? In the entire three-month trading period, there were only 993 unique market participants and this includes players in the credit indices market. Changing terminology slightly, the report states that 500 participants trade corporate CDS at least once a month. Forgive us for scratching our head when these levels of participation are described as “broad.”

Taken together, the market data in the report confirm a description of the CDS market we have long espoused. CDS buyers and sellers are few in number, execute a small number of trades, and prefer those trades to be larger-sized transactions. The CDS market requires strong market-making. Participants use the market because it works and offers the size, price and certainty of execution the cash markets cannot provide. We will pick this up next week and discuss market-making.