A Credit Trader

Bloomberg is out with an article describing how Milan city officials used derivatives to bet on the default of the Italian government. While we wait for the sounds of knee caps being broken, let’s take a deep breath and think about what may be going on here.

The article states that “The city council sold credit-default swaps that protected the banks against a decline in the value of Italian Republic securities”. The immediate thought that comes to mind is that were Italy to default on its debt, Milan would very likely not be able to deliver on its contract. This is far from an original insight – in fact, everyone from Janet Tavakoli (buying protection on Korea from a Korean bank) to Nassim Taleb (buying insurance on the Titanic from someone on the Titanic) have made this exact point. So, what’s going on? Are the banks that are buying Italy protection from Milan just plain dumb?

There are two important things to keep in mind whenever you read any piece of financial news in the press:

1. though often the broader fundamental point may be correct, the actual phrasing is often misleading if not downright wrong,
2. we should not always rush to brand the traders behind these deals as unaware of potential risks

For the ADD-afflicted, my take on what’s going on is in the section entitled: Self-Referenced CDS/CLN

Some Background

At first glance, the article suggests that Milan has ventured close to the world of self-referenced credit derivatives, just the kind of market that would endear itself to those who hate the concept of plain vanilla CDS in the first place.

The question is why would anyone buy protection from a reference entity on itself since the entity in question will be unable to honor its obligation? Or in other words, how can we possibly trust an entity to guarantee its own debt?

To take a step back, we have to ask whether entities are allowed to buy or sell derivatives on their own securities.

Self-Referenced Equity Structures

Employee Options

The most obvious instance of entities trading derivatives on themselves can be found in stock options offered to employees as part of their compensation packages. Outside of backdating scandals and the ongoing accounting treatment issues, these products have been adopted as an accepted part of the market economy. In fact, they are likely to gain popularity, especially in the financial sector, where cash bonuses are likely to give way to options with long vesting periods.

Selling Equity Puts

Many companies, such as Microsoft, have sold equity puts on their own stock as a way to cheapen the cost of stock buy-back programs or lock in their cost. Since the 1992 SEC ruling, these trades have been attractive to corporates, driven partly by the tax-free treatment of the put premia received by the companies. The basic motivation behind the trade is that if the stock price rises, the puts expire worthless and the company offsets the higher stock price it pays on the repurchase (if it decides to go ahead with the repurchase anyway) with the premia it received on the puts. If the stock price falls, the company is essentially delivered its own stock at the strike price – thus having locked in the stock repurchase cost at the time of the sale of the puts.

Self-Referenced Debt Structures

Synthetic Debt Repurchase

Stock repurchase programs are fairly common and well-established. In the fixed income world, the debt equivalent of a stock repurchase plan is a Debt Repurchase which can take the form of either Cash or Synthetic programs. Though less flexible than their synthetic counterparts, cash debt repurchases can be executed via open market debt purchases (which is more common for a buyback of a fraction of the outstanding debt) or via public debt tender offers (which is geared toward buying back all or a significant portion of an outstanding debt security).

Synthetic debt repurchase seeks to replicate the economics of its cash cousin with enhanced flexibility and reduced execution costs. Here, the issuer enters into a Total Return Swap with a bank with a particular issuer’s debt security as the underlying asset. The bank will pay the issuer interest and principal payments made on the underlying security and, to the extent the maturity of the TRS is shorter than the security, settle the price difference at maturity. Similar to the cash outlay on the cash repurchase, the issuer will post collateral on the TRS to the arranging bank.

Self-Referenced CDS/CLN

Here, the company sells protection on itself. Though they are, in effect, identical, there are two flavors of the trade: 1) collateralized CDS or 2) Credit-linked Note. A fully collateralized CDS is essentially a CLN (i.e. a bond)., while a partially collateralized CDS (say, 50%) is a called a leveraged CLN. Such trades usually have embedded spread or MTM triggers that are tied to the either the level of the CDS or the MTM of the trade.

This trade may be appealing to a company if its bond/CDS basis is positive i.e. it earns a higher yield by executing the trade via CDS rather than cash bonds. Also, if the company believes the widening in its credit spreads is overdone, it can sell CDS on itself and unwind at tighter levels, though it would then be open to insider and/or market manipulation charges. Finally, as a funding trade, it may make sense for a corporate to buy its own short-dated CLN at L+x, rather than put cash on deposit or buy Tbills that offer sub-Libor yields.

In fact, this is what MBIA did in 2002 when it bought CLNs linked to its subsidiary. Bill Ackman has talked about this trade in his broader criticism of the company. Effectively, by purchasing its own CLN’s MBIA benefitted publicly by improving the weighted-average rating on its portfolio and causing its CDS spreads to tighten, while at the same time, hurting its liquidity (by having to post cash for the CLNs) and leveraging itself up on its own creditworthiness - two issues that were not disclosed.

As far as the Milan trade, the story looks pretty simple – Milan had some cash that it needed to put to work. Rather than go to the market and buy Italy bonds (which I’m sure is well within the mandate of Italian municipalities just as buying US Treasury bonds is well within the investment mandates of US muni’s), it chose to structure the trade in a synthetic fashion – either to match a particular maturity/target yield/notional schedule etc.

So, a tad overblown if you ask me.

Collateral Enhancement

Though potentially a red flag, a company can, in some cases, increase its credit line with a lender by assigning a long protection trade as collateral, though the increase will not likely be 1:1 for the amount of protection given increased counterparty exposure to the lender on the CDS trade.

Improving Recovery/Funding

A neat trick that improves recovery prospects for subordinated debt holders while providing senior-level funding for the issuer without increasing the outstanding amount of senior debt involves selling sub debt to an investor and at the same time entering into a TRS where the investor pays the total return on the sub debt in exchange for L+x.

This achieves the following benefits in some jurisdictions:

• The senior noteholders see an improved position in the capital structure with the sale of subordinated debt.
• The investor sees an improved recovery profile on the net trade. Say, sub and senior recoveries are 20% and 50% respectively. The investor receives the 20% on his sub debt plus 50% x (1-20%) on the TRS, which beats the recovery of the senior noteholders.
• The issuer obtains senior-level funding, without showing any increase in the outstanding amount of senior debt

Improving Credit Ratios

A way for a bank to improve its credit ratios is to do the following. Find a bank (investor) to put up $100 for issuer’s subordinated debt. This $100 is then roundtripped back to the bank in exchange for a zero-recovery CLN linked to the issuer. So, in case of a default the bank will find itself with a liability worth zero and an asset worth the subordinated recovery, a net of greater than zero. The issuer will be required to hold some $8 of capital against the CLN, with a net capital increase of $92.

Clean Asset Swaps

Let’s say you’re an Emerging Market corporate or a sovereign that would like to issue a sizable amount of debt. You have two choices: you can issue in your local currency or you can issue in a foreign currency, let’s say a major currency like USD. The benefit of issuing in the local currency is that you are not taking any FX risk i.e. your revenue/taxes are in your local currency – the same currency which you will use to pay interest to investors. The downside, however, is that the market for locally-denominated debt may not be large enough to digest the issuance, at least at levels that are attractive to you. On the other hand, issuing in USD taps a much larger market, however you will now have pretty serious FX risk as the size of your debt will grow if the local currency depreciates relative to the USD. This is the problem of “original sin” that all emerging market issuers are keen to avoid.

So, what’s an enterprising Emerging Market issuer to do? The obvious solution is to hedge the FX risk with a cross-currency swap. You issue debt in USD and then enter into a swap where you receive USD from a bank and pay the local currency, thus hedging your exposure. Feeling pretty good about yourself and about to pull the trigger, you get a call from a credit marketer of the same bank who tells you that he can pay you 50bps more on the swap than the rates desk. What’s the catch? The catch is that in the event of default by you, the cross-currency swap is extinguished.

This is not much of a catch – who cares what happens in the event of default, especially if you have one less trade to worry about during the workout/restructure process. An important caveat is that this trick can only work when local currency rates exceed foreign currency (USD in our example) rates. Why? If local currency rates are higher, that means the currency is expected to depreciate relative to USD throughout the trade, which means that while the earlier cashflows are expected to be a net positive for the bank (+ve carry), later cashflows, including the large principal payment, are expected to be worth little for the bank. Adding credit risk to the structure means the scenarios when the bank is receiving cashflows that are worth less than what it is paying are less likely, hence the bank can afford to pay more to the issuer on day one.

This trade has been very popular, for instance, with Latin American sovereigns but less so with corporates, largely due to the lack of liquidity on corporate CDS. In fact, a good guage of whether corporates/sovereigns are doing this trade can be found in the quanto CDS market where banks will go to hedge the local-ccy denominated CDS risk they are taking on via this trade.

Kangaroo Bonds

This structure involves a higher recovery in case of default for the investor in exchange for better financing to the issuer. Say, an investor buys a $100 issuance and at the same time buys $200 of CDS from the issuer. In case of default and 50% recovery, he will recover $50 on the bond and will have a claim of $100 against the client, on which it will recover an addition $50, this making him whole. He will consequently pass on this benefit to the issuer via significantly better funding.

Lawsuits… here we come

Trades are going sour and the municipalities are rebelling…

• Milan Police Seize UBS, JPMorgan, Deutsche Bank Funds
• JPMorgan, 11 others sued over Jefferson county crisis
• Derivatives Hit Austrian Railroad With Record Loss

… and market spreads are responding

implying default rates that illustrate to what extent the market is dislocated.

Find the one that doesn’t fit
Municipal CDS market spreads (via the MCDX index) imply a 5y cumulative default probability of 34% versus a historical default rate of 0.02% (Moody’s) - a figure 1500x higher!

Purists in the room will claim that we cannot rely on a rating agency for historical default rates, as statistics come only from rated issues. Additionally, the default numbers do not take into account conduits where the default rate can run 10x higher – in fact, 2008 alone saw over $6bn of muni defaults – 19x higher than the previous year, though this has been in non-GO and unrated issues all of which recovered at 100% (anyone involved in the CDO market would laugh at a “default” that recovered at 100%, having been inured to the trusty old 0%).

The high market implied default rate is also skewed by the high expected (marked) recovery for muni bonds of 80% versus 40% in high grade (a higher recovery implies a higher default rate for a given CDS spread level) as well as potential uncertainty over deliverable obligations into muni CDS.

Each down cycle in markets exposes those unsophisticated investors who are “swimming naked”. Municipalities tend to be perfect examples as the Lack of Sophistication / Assets Under Management quotient is strikingly high. Though the current crisis is presenting muni’s with more than badly gone trades, it is these trades that offer the biggest lessons to investors.

Below I walk through what has gone wrong and what lessons have been learned (or relearned). In particular, I suggest the following guidelines to both the municipalities that have gone off the trading deep end as well as the banks that have assisted in the process.

Suggested Guidelines (fleshed out below)

Investment guidelines for municipalities:

1. Take no basis risk
2. Use no leverage
3. Make no punts

Risk management guidelines for banks:

1. Do not offer double-down trades
2. Pay attention to “willingness to pay”
3. Use credit mitigants like collateral and mark-to-market trigger agreements

Munis under stress
Though Muni spreads are clearly too high, it is as equally true that many municipalities are under economic stress. In April Moody’s assigned a blanket negative outlook to the entire U.S Local Government sector, citing fiscal challenges as a result of the housing market collapse, dislocations in financial markets, and a deep recession.

• Total fiscal 2009 budget deficits stood at $72bn, or 12% of general fund assets, according to the Center on Budget and Policy Priorities
• Decline in state revenues is accelerating due to a decline in sales and tax receipts
• Tight capital markets are constraining raising funds, though issuance has recovered since the Lehman default
• States are facing challenges with their pension obligations just as falling equity and corporate bond prices have eaten into investment pools

With Jefferson county teetering on the edge of Chapter 9, Vallejo, Calif recently entered Chapter 9 protection. Other California cities, including Rio Vista and Iselton have also said that budget gaps may force them into insolvency.

Outside of the difficult economic environment, munis have also been embroiled in their own financial mess, involving VRDO’s, TOBs as well as the monoline insurance disaster all of which are putting additional pressure on finances. In fact, trading decisions made by municipalities now figure in municipal investor analysis just as strongly as fundamental indicators like unemployment, interest expense and general revenue.

The Brave New World of Muni Trading
Below I list the trades executed by municipalities that have been particularly prominent in the news.

VRDNs
The most interesting trade to come out of the muni crisis has been the debacle seen in the variable rate market. The two types of variable rate muni securities are VRDOs and ARS. They are broadly similar but vary in puttability (VRDO’s are puttable, ARS’s are not), denominations, monoline coverage, investor types etc.

The most important thing, which is common to both types of securities, is that the interest rate resets periodically according to some basic rules (auction-failure for ARS or a result of the remarketing process/changes in specified index for VRDO). The interest rate resets either to whatever rate clears the market or in the case of auction failure, steps up to a very high level so as to compensate those investors who were not able to sell their securities. These securities were very popular with muni issuers simply because, even after all the structuring fees (bank letter-of-credits and monoline insurance), the variable short-term rates paid by issuers were less than if they had issued long-term fixed rate debt.

The rates paid on the VRDO’s have tended to track the front-end cash/Libor levels. For this reason, the municipalities, bowing to the wisdom of risk management, hedged their interest rate exposure by paying fixed on interest rate swaps (or buying interest-rate caps). The floating index on the swaps was typically 67% of Libor. So, in the ideal case, the muni is paying a floating rate while receiving pretty much the same rate on its swap hedge on which it pays fixed.

When the market grew suspicious of monolines, paricipants rushed to put the bonds back to the market just as the bids retreated from auctions. Clearing rate levels shot up and ARS’s hit their maximum rates – most well-known in the case of Port Authority which started paying 20% vs. an earlier 4%.

So, the net result was that the floating rates the muni’s were paying on their VRDN’s shot up (owing to general illiquidity as well as the concern around monolines), while the floating rates they were receiving on the swaps collapsed (as interest rates fell) – meaning the muni’s managed to lose money both on their bonds as well as the hedges.

In retrospect, the municipalities made a number of key mistakes:

• Liquidity risk – the most fundamental mistake was similar to the one that befell the Structured Investment Vehicles: the reliance on short-term debt. Once cracks appeared in the monoline wraps and investors withdrew into cash, the muni’s were not able to roll over their debt. The municipalities were effectively subsidizing the rate on their debt by writing a massive liquidity option. They compounded the problem by backstopping their own short-term debt with impossibly high rates, knowing full well that paying such rates on their debt was not sustainable. The realization by the market of lack of demand for variable-rate securities and the unstainably high rates paid on the debt only added to the lack of confidence in muni debt
• Monolines – anything having to do with monoline wraps tends to be procyclical meaning it will exaggerate the pain on the wrapped products. The basic point is that the nature of the monoline business suggested that the time when an investor will seek to benefit from the wrap, is exactly time when that wrap will be worthless. Apparently, some municipalities have also agreed to post collateral in the case of ratings downgrades, something which would tend to happen when the monolines are gone, the economy is in recession and cash is at a premium and difficult to raise and swaps are negative-MTM to the muni’s.
• Rates – municipalities may be forgiven for failing to foresee the silly price action in 30y swap rates which are still trading below 30y treasury yields. This was due to hedging of structured rate notes by dealers issued mostly to individual investors. The massive rally in rates has led banks to receive fixed in size in the long end of the swap curve in a largely illiquid market. This has led to large losses to the swap hedges done by municipalities. Most of these have tended to be in the long end of the curve, just where the rally has been most brutal.

Selling swaptions
Second on the list of poorly-thought-out trades has been the selling of interest rate swaptions. On the face of it, the motivation for selling swaptions looks quite reasonable. The muni issuer sells a payer swaption (on exercise the muni will pay the fixed strike). If the swaption is exercised, the muni pays fixed, receives floating and at the same time issues floating-rate bonds and uses the proceeds to refund the outstanding issue of bonds. Net result is a locked in synthetic rate for the issuer with some savings from the swaption premium.

This all sounds well and good except for the fact that the banks end up dictating if and when the municipalities issue debt. This decision to issue debt is not based on any fundamental funding need, but rather on the performance of a single trade done with the bank.

A cursory look through the news shows that Erie School District, Philadephia National Airport and Alabama Schools have all sold swaptions. This suggests that if rates were to fall they, and many other counties and districts who have also sold swaptions would need to come to the market and issue debt, leading to potential supply dislocations.

Structured Notes
This round of municipal crises has been (yet?) notably clear of structured note exposure, unlike the Orange County episode. Recall that Robert Citron put a large part of the portfolio into structured notes, particularly, inverse floaters. In fact, the county bought more inverse floaters (in notional terms) than there was equity in the fund. In the end, it was the leverage (about 2.6 at the highs), financed via reverse repos, that largely led to the $2bn of losses for the fund.

The particular danger in inverse floaters has to do with their duration profile. As inverse floaters are floating-rate products (the rate is linked to the level of interest rates), one would be forgiven in thinking they have short duration. In fact, the duration is higher than for a similar-tenor fixed-rate bond and increases as the trade goes against the investor (i.e. it starts behaving more like a zero-coupon bond).

What blew up Citron was the fact that he took on two kinds of leverage (you could say, he tripled-down): 1) the particular investment that he chose to express his views – the inverse floater was already dollar-for-dollar a leveraged investment as the duration of the product increased precisely when it went against Citron and 2) Citron put more money into inverse floaters than there was equity in the fund.

In addition to leverage and unappealing duration risk, the county had credit exposure to the dealers which was less of a concern then but is clearly much more significant now. In fact, given the current experience, real money investors are much more likely to structure trades with SPV’s rather than dealers directly in the future.

FX KIKOs
FX KIKO trades are the latest in the long series of toxic trades executed by unwitting investors during the latest market cycle. In fact, they have been so popular that they’ve been described in Bloomberg magazine, where they were called “I-kill-you-laters”, rather than “Accumulators” – their traditional pre-crisis name.

Though rumours of potential pain by European municipalities is only beginning, recent victims include corporates like Gruma – the world’s leading tortilla manufacturer with $700mm losses on the Mexican peso, Citic - with losses of $2.7bn on the Aussie dollar as well as countless other companies and investment houses. Although they come in different variations, these are essentially carry trades with some additional bells and whistles (or, if you like, smoke and mirrors).

Without going into detail, the basic structure of the trade can be summarized by the following chart.

Salient points of this trade are the following:

• If the market goes in the client’s favor, they make a little money and the trade matures in short order
• If the market goes against the client, the notional amount on the trade increases and the maturity extends.

Exporters who did the trades claimed they were being done for hedging purposes. For example, a Mexican company exporting tortillas to the US market received dollars and needed a product that effectively made them short USD/MXN (ie whereby they sold dollars and bought pesos in the market). It is true that these trades positioned these companies the right way, unfortunately the trades were done in much larger size than the hedging operations required and the timing (knock-out components) plus leverage on the trades (which typically flipped from 1 to 2x the size when it went against the client) had no fundamental bearing on the actual financial flows.

It’s not clear how popular this trade was with municipalities though history of municipal involvement with FX stretches all the way back to 1995 when the State of Wisconsin Investment Board lost $95mm from MXN peso trades. Add to this the fact that municipalities pile in, along with companies and retail investors, into popular trades at the top of the market suggests that should see losses on such trades may be disclosed soon enough.

Categories
The classification of the trades described above can be summarized as follows:

1. Basis Risk – these are trades done for fundamental reasons, like obtaining cheaper funding, but go sour occasionally because they expose the investor to some tail risk. VDRN’s was one such trade that exposed muni’s to tail funding/liquidity risk as described above. The key here is to understand that the only reason muni’s were cheapening their funding was by selling a liquidity option to the market. So long that these tail options are viewed as risk free, municipalities will continue making the same mistakes. In reality, nothing is free in the market, and issuance decisions will be made on a more sound basis in the future if all risks are properly taken into account rather than swept under the rug.
2. Leverage – these are the typical “investment” trades based on a particular view of the market. The danger lies in the execution of the view (e.g. structured notes), as well as the leverage taken separately by the investors (via reverse repo’s).
3. Punts – these are trades done for apparently “investment” or “hedging” reasons. What differentiates punts from normal investment trades is the fact that the punt is done in a market where the investor can have no competitive advantage or insight relative to other players in the market. In other words, the answer to Ken Fisher’s question “What do I know that others don’t?” cannot be answered satisfactorily. These trades often figure in FX markets and often executed long after the popular press begins writing about their virtues. The FX KIKO, which is based on the carry trade, is an example of such a trade.

Conclusion
In order for the market to move beyond the news cycle of losses by municipalities and other unsophisticated real money investors, there need to be some fundamental changes. These changes have to come from both sides of the equation: municipality investment mandates and bank risk management policies.

Recent news has made it painfully clear that municipalities need to abstain from certain kinds of risks. These risks include the 3 categories mentioned above because each one has the potential to bury them, if left unchecked. In general, municipalities do not have the resources to run a full-fledged asset management organization, suggesting that their investment mandates need to shrink significantly. Although, generally speaking, the allocation of more power and decision-making ability to the local levels is not a bad thing, what we cannot have is a situation where part-time town board members without strong and dedicated risk management support staff make decisions that put at risk pensions of thousands of other people, while guided by biased investment advisors.

Investment mandates for municipalities need to be made on a state, if not the federal level and there has to exist a framework for escalating problems higher up before they explode. For example, a county finding itself in financial difficulties should not think that the only way out of trouble is to finance its budget by selling a ton of swaptions to the next friendly dealer that comes along.

On the bank side, dealers need to adapt consistent trading and “know-your-client” standards with municipal counterparties. These should not only include a relatively conservative set of products, but also collateral calls so that “punts” and leveraged trades, to the extent they are allowed to happen at all, are unwound/collateralized early and are treated with more care and tracked closely.

The traditional treatment of counterparty risk does not hold for unsophisticated investors like municipalities who can easily claim in court (often with good reason) that they did not understand the risks in the trades they were doing. Clearly, this will happen only when trades go bad adding its own particular wrong-way risk to these trades for banks.

There needs to be greater emphasis placed on the motivation behind client decisions to do certain trades. For example, an interest rate position done for genuine hedging of risk (preferrably without any basis risk) is very different from a short swaption position done to raise cash. The latter is a clear red flag and suggests that the client is essentially doubling down to get out of a financial jam. The risk management function of the banks should ensure that such trades are not done.

In late 2007, as the monoline act of the crisis drama played out, credit traders were being repeatedly pulled off the desk to attend two kinds of meetings. In both they heard bad news.

CDO Writedowns
In the first type of meetings, heads of trading desks would say they were writing down their exposure to a particular monoline. For instance, in the case of ACA, the ugly stepchild of the monoline business, Merrill wrote down $3.1bn in exposure, CIBC wrote down $2, Calyon $1.7bn and Citi $900mm in the fourth quarter of 2007 when ACA was downgraded from A to CCC. Prior to that, the banks had reported having little net exposure to CDO’s as their long bond positions were matched by CDS hedges.

The problem was, of course, that the hedges, or short positions, were done mostly with monolines. Merril, for example, gave their net CDO exposure only as $7bn, which consisted of $30bn gross CDO exposure against $23bn of hedges. What the bank omitted to say was that $20bn of those hedges were on with monolines. Assuming, 50% losses on the CDOs and a default by the monoline, real exposure was actually more than double the declared amount.

Lehman

Estimates of net CDO protection written by monolines come to around $125bn of mostly super-senior but also some junior super-senior and mezz exposure. Assuming 40%/60% recoveries, losses from monolines defaults would come out to around $50bn. This is a conservative assumption as

1. not all monolines would default
2. the banks hold some collateral / have collateral agreements in place (AAA monolines had much less strict collateral posting provisions and it appears even ACA, which due to its A rating was required to post collateral, won forbearance agreements from its counterparties)
3. the banks hold hedges against the monolines (though this will have been recognized long before the writedowns on the cash assets as vanilla CDS on liquid names are marked-to-market)

Without going into detail, the case of Merrill is particularly disheartening from a risk management perspective as it appears to have committed two basic mistakes. Just as the investors were becoming wary of CDO’s, and ABS CDO’s in particular given the apparent weakness in the housing market, one division of Merrill continued accumulating cash ABS assets just as another division was failing to find buyers for the securitized products.

Why Merrill was aggresively buying up assets it knew could no longer be offloaded is puzzling and speaks of disincentives in the firm. At some point though, the bank did realize that this practice was not a particularly wise strategy and if it couldn’t find buyers of the cash assets it was going to find a seller of CDO protection. At that point the only insurer willing to stick its neck out was ACA (XL Capital having walked away) which was even then on particularly shaky ground. Merril put on $10bn of hedges with the firm, however, soon after ACA was downgraded and the bank was again swimming naked.

The obvious question is why did banks have exposure to monolines in the first place? The short answer is the negative basis trade (isn’t it always?).

Moody's

The slightly longer answer is that, as mentioned, above some banks used the monolines to hedge their exposure to CDO tranches while they waited to offload them to investors.

1. For example, say a bank comes out with a $1.5bn issue of a CDO but can only find interest for half the size. The internal trading desk, whether it wanted or not, would have to hold whatever wasn’t placed.
2. Another reason was that a monoline wrapped tranche could be marketed as a higher quality product (super-AAA rather than a plain AAA) and so could reach a broader set of investors.
3. Also, buying protection from monolines was often the only way to manage the risk of CDO assets as single-name CDS on ABS CDOs hadn’t come into the existence (or at least standardized existence) yet.
4. Finally, this trade was attractive from a regulatory capital perspective and, more importantly, was positive carry which meant that holding it on the books seemed like a good strategy given the cheap balance sheet cost and a failure to recognize that funding costs may rise in the future.

Monoline Risk
The second type of meetings that credit traders were ushered into were with internal sales teams who marketed wrapped products to the banks’ clients. These were typically municipal bonds, including the poster child of muni distress: auction-rate securities. Prior to the monoline crisis, banks and investors were not particularly concerned with proper valuation of the credit risk in these securities as the monolines stood ready to pay up if the issuer were unable to do so.

However, with the monolines faring far worse than the issuers whose credit risk they were guaranteeing, investors began discounting the wrap and focusing on the underlying risk in the products. Suddenly, salespeople, whose eyes would glaze over at any mention of credit risk, were forced to understand first and second-to-default basket products and the concept of default correlation.

In a way, these “vanilla” folks had a better sense of what was happening than the more “exotic” CDO originators. They understood the fact that the guarantees offered by monoline wraps were largely illusory and offered no protection or diversification of risk.

Prior to the crisis, monoline wraps were viewed as monoline risk. This leap of faith required two assumptions: first, that the rating of the monoline was higher than the rating of the product it was insuring – this seems commonsense (ignoring for the moment the controversy over artificially low muni ratings), however this is an assumption that went out the window in the case of ACA-insured CDO tranches. The second assumption was that the default risks of the monoline and the underlying product were uncorrelated. I go into this in more detail below, but siffice it to say that in the extreme case of perfect correlation between the two entities, the monoline is expected to default at the same time as the underlying product, suggesting that the wrap added no extra protection. If 100% correlation seems high, consider the fact that the monolines were thinly capitalized relative to the risk they guaranteed (something that was made clear after the fact) and that both the monoline and the product were fundamentally exposed to systemic risk, a scenario in which both would and did suffer massive losses.

The Monolines’ Way-ward Ways
In my AIG post, I’ve described the concept known as “way-ness” which, essentially, requires us to consider the likely state of the world in the scenario a given entity of product defaults. This is particularly relevant in managing counterparty risk. For example, all else equal, you would much rather have an airline client sell you puts on WTI than calls. This is because a lower oil price (though not too low) is more likely to boost airline profits and the company would have less difficulty in making good on the put contracts.

What are the considerations for monoline way-ness risks:

1. An important question a bank has to ask itself each time it does a trade with a client that may expose it to counterparty risk (i.e. a derivative rather than a fully funded trade) is what is the client’s motivation for doing the trade. For example, if a client is punting in the market in an attempt to make up for heavy losses on other trades, the bank should be more cautious. In recent years, municipal insured penetration has steadily declined, driven by increased investor risk appetite (less need for wraps) and a perception by municipalities of a fundamental bias in the rating methodology for their sector. This has led the monolines to expand into new markets and consider new business opportunities just as the price of risk was hitting new lows. Rating agencies were also keen to expand the structured products business for which they gathered high fees. They may also have influenced the monolines to pursue the business as a way to diversify their revenue stream and ultimately keep their AAA ratings. Also, a client that is aggresively pursuing highly leveraged opportunities outside of its historic mandate, for example a corporate putting on exotic curency trades, is particularly at risk as it points to possibly broken risk controls and poor risk management. As the market discovered later, the monolines that were particularly aggressive in bidding for structured finance business, such as ACA, were the ones that would be less able to withstand losses in a stress environment



2. The key difference between the staid muni bond insurance business and the new CDO tranche insurance business that monolines were underwriting has to do with  the static/dynamic mark-to-market risk profiles of the two products. This is an important consideration as the mark-to-market gains and losses are linked to the capital the insurer is required to hold as well as the collateral it may be required to post. An insurance provider prefers a lower mark-to-market sensitivity (and hence less onerous capital charges and potential collateral calls) if the product it is insuring performs badly, all else equal. For example, for a typical bond, the MTM sensitivity decreases as the credit risk worsens (since the duration of the bond decreases) – exactly what the insurer prefers. In the case of a super-senior tranche, however, the MTM sensitivity will actually increase – precisely in the worst possible time (since the delta of the tranche will increase). To provide intuition for this, consider the tranche as a initially deep-out-of-the-money option on expected loss. The wider spreads rise, the closer the super-senior tranche is to suffering impairment and hence the higher its sensitivity to credit spreads. So, as the performance on these tranches suffers, the insurers would be marginally more on the hook for each basis point wider in spread. An attempt by the insurer to delta-hedge its exposure is likely to lead to losses since the monoline would be hedging a negative-gamma position and locking in losses with each rebalancing trade as spreads move.
3. If at some point, monolines decided to offload their super-senior tranche exposure because of their view of the market or as a preventative measure to shed risk, they would likely find it very difficult to do so. The scenario in which super-senior tranches are under stress is a scenario when liquidity is at a premium and buyers of risk are on the sidelines.
4. Going back to the point briefly mentioned above, super-senior tranches are likely to sufer in a “systemic” crisis environment – precisely the one in which we find outselves today. This environment is the one where the monoline itself would be struggling, unable to source new business as issuance and risk appetite dry up. It would be difficult for the monoline to find new sources of revenue to offset the bleeding in cash due to collateral postings or recapitalization efforts.
5. Poor performance of super-senior tranches would put additional stress on the monolines which may ultimately lead to ratings downgrades – precisely what we have seen in the last few years. Ratings downgrades would automatically trigger collateral calls and increased capital cushions, precisely when the monolines would be least able to afford it. Though rating agencies try to rate companies “through the business cycle”, the fact is that there are more downgrades than upgrades during recessions.
6. The hedging of monoline exposure by banks will increase the stress on these companies. Since the monoline spreads are correlated to super-senior tranches, banks having counteparty exposure to these insurers found that their exposure increased as credit spreads widened. This caused them (at least for the one who were actively hedging their exposure) to buy protection on the monolines exacerbating the perception of monoline risk in the market and leading to a self-fulfilling spiral.
7. Recoveries in an environment of high defaults would be lower than average putting further pressure on monolines.
8. An alternative to buying protection on the monolines to manage the banks’ exposure was to buy protection on the underlying bonds in the CDO. This was not possible in the early stages of the market as single-name ABS CDOs did not trade. However, in the last few years liquidity improved and banks were able to source protection. However, buying protection on CDS will likely push cash bond and CDO spreads wider which will increase banks’ exposure and cause further MTM losses to the monolines
9. We should differentiate willingness from ability to pay. Insurance companies are generally loath to make payouts on policies, so it is not surprising that we would witness monolines balking at making payouts on the CDO tranches. Merrill, sued SCA over $3bn of CDS positions. AIG has also tried to wiggle out of some protection it wrote. This is not surprising as none of the insurers ever considered it remotely possible to have to pay out on these contracts which contributed to insufficient reserves and lax risk management. There is also some controversy over side letters suggesting that neither the insurer nor the insured expected cash to change hands on these contracts.
10. Though few people take ratings very seriously (especially now), the fact was that Merril entered into contracts with ACA (then A-rated) to buy protection on a AAA underlying. The ratings suggest that ACA is more likely to default than the product on which it is providing insurance. This is actually worse than “buying insurance on the Titanic from someone on the Titanic”.

The monoline crisis provides a textbook case for the do’s and don’ts of managing counterparty risk and why some banks ended up suffering much more than if they had pursued a more sound risk management strategy. Greed will get you every time…

Pop-quiz! In which market can you actually lose money by being too right? Well, credit derivatives, where else? Read on for an explanation.

One of the recent Big Bang changes in the CDS market has been the recouponing of CDS to 100bp or 500bp. This change achieves two things: enhances fungibility and liquidity for clients and solves the vexing risky annuity issue that has brought much pain to dealers throughout the years.

Why Recoupon CDS?
To illustrate the basic motivation for the recouponing change consider a reference entity that has one bond outstanding and no loans. Trading the credit of this company is pretty easy: you’ve got a single security with a fixed coupon and an observable market price.

Now imagine that CDS starts trading on the name. Each day the spread moves you’ve got a different tradable product out there whether at 100bps or 150bps or 500bps or even upfront + 500bps running. So if a dealer does 10 bond trades, he’s either paying or receiving the coupon on the bond so the trades net out, nice and simple. However, if the same dealer does 10 CDS trades, it will likely have 10 trades all with different coupons and hence with different time decay, convexity and duration profiles, making risk management more difficult. Here, by the way, we are not even considering potentially different restructuring clauses, lookback credit event effective dates, roll dates and accrual periods.

Why not simply unwind previous trades, rather than put on new trades? Well, unwinding previous trades essentially means trading in off-market CDS which, as first time CDS participants quickly find out, carries an extra bid/offer cost – in fact, when quoting a level many dealers will explicitly ask you whether it’s for a new trade or an unwind of an existing position. The other problem is that only clients (i.e. buyside firms) can unwind CDS with dealers, not the other way around. In other words, dealers cannot exactly call up a client and ask them to unwind CDS trades just because it suits the dealer’s book.

So, if dealers charge extra for unwinding existing CDS positions, why don’t clients just put on new trades? The quick answer is that the practice guarantees continuous growth of the CDS book which will significantly stretch risk management/systems resources of small funds. The extra complexity comes not only from booking the trades, but also from calculating daily exposure and  managing collateral on the positions. Also, by doing new trades, rather than unwinding existing positions, a fund will add counterparty risk, which it may need to manage by buying protection which adds even more trades to the book.

Enhanced Fungibility
Moving CDS to fixed coupons will make CDS trades more fungible and increase liquidity by making all trades “on-market”. This is all well and good and ensures that clients are happy as they will no longer be ripped off trying to unwind existing positions with dealers or suffering operational burden from the explosion of the trading book or having to manage dealer counterparty risk. These are all solid reasons for the recouponing. However, I would argue that the biggest beneficiary of the recouponing will be dealers as the change will eliminate the dreaded risky annuity risk which has been the bane of CDS risk management for a long time. What is risky annuity risk?

Risky Annuity Risk
To illustrate risky annuity risk, consider the following. Let’s say a dealer bought protection on a name at 100bp. The name subsequently widens and the dealer does a new trade selling protection at 300bp. From the looks of it, the trader should be happy as he is now flat the name and has booked a profit of 200bp running. Assuming $50mm notionals on each leg and a risky duration of 4.5, the trader is up $4.5mm on the trade (a simple present value calculation of 200bp for 5 years) – off to the bar with the lads!

A month goes by and the trader is focused on other things. A headline flashes across his bloomberg screen – the company in question has just filed for bankruptcy. Given that he is “flat” the name, he shouldn’t care. Yet he does – in fact, he has just lost $4.5mm. The problem, of course, is that the 200bp annuity he is receiving is risky to the survival of the company. If the reference entity suffers a credit event, both CDS are triggered and the coupon streams go away.

This is why it isn’t always fun being the dealer and making bid/offer on CDS. Clients can come back to the dealer and unwind existing trades (if at a higher cost than new on-market trades). Dealers do not have this luxury – and their books are spider webs of risky annuities with very significant jump-to-default risks that are essentially unhedgeable. The heads of credit desks did not have a happy time having to explain to CEO’s why they periodically lost tens of millions of dollars on credits that they were flat or even short. CEO’s quickly got up to speed with the fact that you can be short delta but long jump-to-default risk.

In fact, this was such a pressing problem that several dealers have attempted to create products to manage this risk. One of these products was an Annuity CDS which essentially turned a risky annuity into an upfront payment. However, since these products would never be as liquid as standard CDS and that they essentially solved a problem for dealers without offering any particularly compelling features to clients, they failed to take off.

Recouponing CDS is the right, if belated, step in the right direction for the market which should make both the buy and sellside a touch happier in these otherwise difficult times.

“Taken out to the woodshed and shot” is a phrase you often hear from CDS traders when one of their names blows up. The same can easily apply for recovery rates in the last few months across the credit markets.

High grade data: 2007 had 1 default and was taken out of the sample

This year’s trend is falling:

Lehman
When Lehman Brothers went bust, its bonds were trading in the low 20s, which for a market used to 40% recoveries (and recovery marks in the 50s for Financials) was shockingly low. A saving grace was that the bonds were expected to trade up as typically happens, especially in cases when CDS notional outweights outstanding bond principal which was true in this case. FT estimated that CDS notional was $400bn vs. $127bn of the bonds. (It’s not clear what seniorities went into this estimate or whether FT took other deliverable obligations i.e. loans into account).

That bonds are expected to trade up post a credit event was a precedent largely formed around the Delphi credit event in late 2005 when bonds hit a local low exactly around the weekend bankruptcy filing and then traded up. This effect was believed to be due to buyers of protection seeking bonds to deliver into the credit event settlement. In reality, you could, theoretically, settle a credit event with a single bond (just passing the fax back and forth). The price action was also possibly due to the “buy the rumour, sell the fact” phenomenon, when the bankruptcy was priced in and the actual filing left the field open to distressed players who saw value in the assets.

What actually happened, unlike in the case of Delphi, was that Lehman bonds started falling in price. And they didn’t stop until the credit event auction held about a month later. The recovery at the auction was 8.625%.

Reasons for Falling Recoveries

Excessive subordination: A top heavy optimized capital structure of many companies that favored loans over bonds will cause senior unsecured recovery to fall dramatically.

Liquidation possibilities: Many consumer sensitive sectors have witnessed large declines in earnings that have left them with high leverage. Normally such firms would be valued as a going concern however those that started out the cycle with high leverage (such as the 2006 private equity vintage) may face difficulty obtaining DIP financing which increases the chance of their liquidation.

Supply: There is a strong correlation between default rates and recoveries which is likely a symptom of excessive leverage but also of large supply of distressed paper. With many investors still sitting on the sidelines, a flood of distressed debt will not find a strong bid.

Recovery Valuation

Default rates
To get a rough idea of market-wide recovery expectations, we can use the graph above. Current expectation of the default rate maps to a mid 20’s recovery, which is about 2 standard deviations below the historic average.

Recovery locks
A recovery lock is a tradeable product that allows investors to isolate and monetize their views on recoveries. It consists of two credit default swaps: one vanilla (i.e. floating recovery) and one with a fixed recovery. Depending on his view, the investor buys the protection on one and sells the protection on the other. Upon a credit event, both CDS are triggered and the resulting cashflow is the difference between the actual and fixed recoveries. You don’t have to wait for a default to make money on the trade, as a move in the recovery market will lead to P/L on the position.

Citi

Recovery locks tend to trade in distressed names at 5-10% bid/offer in the 5y tenor. Liquidity is increasing and markets are being made in 30+ names.

Recovery locks nicely dovetail into DDS or digital default swaps which pay out a fixed unit of $1 upon a credit event. There has been a consistent push to replace vanilla CDS by DDS whch will allow the market to essentially trade the probability of default rather than both probability of default and recovery. It will also make unwinds much more transparent as there will be no argument about which recovery to use. The problem with DDS is that the product is not a perfect hedge for a bond position as a recovery view will need to be made and translated into the notional of DDS.

CDS spreads and expected default rate
In the valuation of a CDS, there is a relationship between three variables: spread, default rate and recovery. Any two will give you a result for the third. Though normally, the traded spread and marked recoveries are used to come up with the expectation of default, the relationship can be used in the other direction given a fundamental view on the default rate and market traded spreads giving us the expected recovery.

Capital Structure Arbitrage
In some sectors, especially in Financials, CDS trades across the capital structure. We can use the fact that the probability of default is the same for senior unsecured and subordinated bonds, due to cross-default provisions, plus an estimate of the recovery of one of the CDS in one part of the capital structure to find the market expectation of the recovery of another part of the capital structure. This is the relationship:

Odds-and-ends: Jurisdiction, Sectors and Ratings

Europe has tended to have lower recoveries than US due to the fact that companies are normally liquidated rather than allowed to restructure as under the Chapter 11 process, though this has recently changed. The threat of liquidation means there is less of a premium for the probability the company comes out of bankruptcy. Also, a company that is more likely to be liquidated is going to try harder to milk its assets to the last drop to survive which will result in lower recovery.

Sectors have tended to have different historic recoveries with Utilities the highest (as these companies own hard assets) and Telecoms the lowest.

There is also a correlation between the rating of a company and its subsequent recovery as a slow bleed of a company’s assets is more likely to result in downgrades as well as lower recoveries.

Low recoveries is just another symptom of the excess leverage and stress in the credit markets. Equity tranches held by banks are likely to have been completely written down by now. The danger is that investors in the mezz and senior parts of the ABS/Corporate capital structures will take an actual hit (rather than an MTM hit) that will cause severe losses within pensions funds, insurance companies and municipalities.

In his latest WSJ post, Soros outlines why he thinks “naked shorting” i.e. buying of CDS without a bond position should be banned.

My commentary to his points are in italics below:

• AIG failed because it sold large amounts of credit default swaps (CDS) without properly offsetting or covering their positions. Perhaps I’m reading in too much into this sentence but to expect AIG to somehow have hedged or offset its CDS trades is akin to an insurance company kidnapping sick people who have bought life insurance and sticking them in incubators to prolong their life. AIG sold protection because it viewed selling CDS as an insurance business. The oft-uttered phrase that AIG was a hedge fund are missing the point that these trades were buy-and-hold; AIG was not in the business, unlike a hedge fund, of dynamically trading in the market.
• What we must take away from this is that CDS are toxic instruments whose use ought to be strictly regulated. I like when the conclusion is stated upfront without any salient points.
• It [heavily regulating CDS] would also save the U.S. Treasury a lot of money by reducing the loss on AIG’s outstanding positions without abrogating any contracts. In fact, the US Treasury had three options in dealing with AIG’s trades: 1) take over AIG and have AIG’s counterparties face the government, 2) post enough cash to cover AIG’s collateral calls, 3) unwind AIG’s trades. It chose 3 which I think is the worst option as a) it locks in massive losses, b) it does so at the absolute wides of the market (spreads naturally blew out as soon as the market realized AIG was in big trouble), c) it commits the most amount of cash upfront.
• Since they [CDS] are tradable instruments, they became bear-market warrants for speculating on deteriorating conditions in a company or country. CDS can be used as easily to go long risk as short risk. In fact, for each nefarious speculator betting on the demise of the poor company by buying protection, there is an avenging angel who is sitting on the other side of the trade and is a seller of protection. While, it is true that there can be heavy one-way flow in CDS on the back of strong protection buying or selling which will drive the market in one direction, the dealers obviously adjust the CDS levels up or down based on this flow at which they are happy to take the other side of the trade.
• Thus, we must understand financial markets through a new paradigm which recognizes that they always provide a biased view of the future, and that the distortion of prices in financial markets may affect the underlying reality that those prices are supposed to reflect. “Reflexivity” strikes again. I don’t think it has ever been news that the prices of assets affect investor psychology which will, in turn, have an effect on the prices of financial assets. This is certainly true of all other assets including CDS.
• Going short on bonds by buying a CDS contract carries limited risk but almost unlimited profit potential. By contrast, selling CDS offers limited profits but practically unlimited risks. This asymmetry encourages speculating on the short side, which in turn exerts a downward pressure on the underlying bonds. If the CDS product payoff profile is so skewed in favor of buyers of protection, why did credit spreads rally for many years until 2008. Also, if the average price of a high yield bond is in the 50s vs. an average historic recovery (yes recoveries in this cycle will be lower) of 40 – that suggests that the payoff profile is in favor of protection sellers, not buyers. Finally, gamma is on the side of protection sellers as well as duration increases as spreads rally – in other words, a protection buyer makes less marginal dollars for each basis point of widening in spreads since risky duration goes down. This can be seen via the Merton debt/equity model as well. Also, for distressed names, protection tends to be priced upfront which means that buying protection in expectation of a quick default is actually quite expensive.
• People buy them not because they expect an eventual default, but because they expect the CDS to appreciate in response to adverse developments. I don’t understand what is wrong with this. If the risk of default increases, protection should be more expensive. That’s called a fair market.
• AIG thought it was selling insurance on bonds, and as such, they considered CDS outrageously overpriced. In fact, it was selling bear-market warrants and it severely underestimated the risk. AIG was insuring mark-to-market risk rather than default risk which is where they went wrong.
• A decline in their share and bond prices can increase their financing costs. That means that bear raids on financial institutions can be self-validating. Lehman went bust largely because it could not raise short-term funding, not because of any CDS pressure. A rating downgrade caused the stock price to fall, making it difficult for Lehman to raise enough cash by issuing equity which caused rating agencies to downgrade it further, leading…. Also, Morgan Stanley CDS traded wider than Bear or Lehman and yet it miraculously survived. I guess reflexivity is only invoked when it works, kind of like those technical indicators.
• I believe that they [CDS] are toxic and should only be allowed to be used by those who own the bonds, not by others who want to speculate against countries or companies. What is wrong with speculation as such? Should we ban short-selling in stocks forever and ever? Metalgesellschaft lost a lot of money on commodities and Orange county on moves in interest rates. Let’s ban those as well. Also, reading this sentence suggests that CDS can only be by those who hold bonds. So, will sellers of protection be required to hold bonds as well. I’m sure this is not what Soros meant, just thought I’d be cheeky.

Outside of penning diatribes against AIG bonuses, blaming CDS for the current crisis has been the most popular topic of late. Perhaps we can ascribe this to the fact that “credit” often goes alongside “crisis” in the press or the fact that AIG was caught with its pants down writing worthless protection, I’m not sure. In any case, I think it’s gone a touch too far.

For fear of becoming yet another CDS pundit, I will (try to) keep this brief and return as quickly as possible to interesting things in the credit markets like the current recovery regime as well as credit/equity relative value. (Please feel free to suggest topics in comments or via email).

Mind you I am not a leave-the-CDS-market-be zealot; I do think it needs changes. In particular, we need a way to deal with counterparty risk. We also need to standardize contracts to ensure liquidity and fungibility (this includes restructuring clauses, fixed coupons, hardwired cash settlements etc.). Both of these issues will be covered by the establishment of the CDS Clearinghouse.

Morgan Stanley

Now let’s run through some things that have been mentioned in the press/blogs about CDS.

But CDS is outstanding is $55trn – that’s equivalent to world GDP! – Bill Gross

Yes and the outstanding notional of Interest Rate Swaps is over $300trn, should we ban that as well? The $55trn figure, of course, ignores both the netting of risk (according to ISDA, after offsetting exposures, the true risk is 3% of the headline $55trn number) as well as the recovery (historic average of 40%) .

Now, we can argue that $10mm notional in IRS is not equivalent to $10mm notional in CDS given the different nature of tail risks in the two products, though here I would point to some emerging market countries that have had short-dated rates north of 100% meaning the exposure on a IRS could actually be greated than in a CDS.

But CDS was designed to disperse risk. Now we realize that it was, in fact, concentrated – Gillian Tett

CDS was not designed to disperse risk, per se. Instead, It was designed to do two things:

1. allow banks to get regulatory capital relief on their loan books letting them free up capital
2. allow banks to more efficiently manage credit risk (without CDS, a bank would have to sell the loans to get rid of the exposure, which is something it would be loath to do as the reference entity is likely to discover this, putting the banking relationship at risk).

But CDS contract language is complex and credit event settlements may not work as advertised – Satyajit Das

First, CDS language is actually not that complex if you’ve read the ISDA. Second, if the language is “complex” then it’s because CDS deal with low-frequency and potentially large contingent liabilities which are important to get right. In fact, I would use the word rigorous rather than complex. As far as auction settlements, witness the 30+ smooth instances in the last three years, especially in the case of Lehman and the agencies.

On the second point, Das uses the example of the Delphi settlement writing “Delphi had 37% recovery when recovery was set by Fitch at 1-10%”. This is very misleading. The 37% recovery was a level where a) Delphi bonds were actually trading at the time of the settlement and b) where holders of bonds could be made whole against their protection positions (in any case the holder of bonds and CDS does not care where the auction recovery is settled since the P/L on the bond is offset by the CDS regardless of the actual recovery).

My guess is that the Fitch recovery numbers was a fundamental view of where the recovery would be on the bonds had Delphi gone through the workout process. This number really has no bearing on CDS (as CDS is not designed to hedge against the final workout price), especially if your view is that Delphi intends to come out of bankruptcy protection which tends to be the case with American companies.

But AIG failed because of CDS

This one is pretty hard to argue with. Yes AIG wrote massive amounts of protection on supersenior tranches of ABS CDOs. As spreads widened, it had to post increasing amounts of collateral. Further, a downgrade triggered ratings-based collateral triggers which quickly led to its demise.

Here I would argue that it wasn’t CDS, as such, that led to the failure of AIG. Rather it was the regulatory/ratings/trading environment of CDS.

First, AIG never had to post collateral when it entered the trades, something which led them to view the business as “free money” and likely caused them to sell more protection than they would have otherwise.

Second, collateral postings were not managed well as AIG had continuous disagreements / negotiations with its counterparties on the amounts to post. Presumably, if they were required to post collateral daily they would have acted sooner to unwind their positions.

Third, ratings-based triggers exacerbated the problem as such triggers are procyclical and subjective.

I would also argue that CDS should not be considered by insurance companies as a business opportunity. I and others have commented on the high correlations (both between individual bonds in the CDO as well as performance of CDO’s in an extreme event). I will just add two points that further draw a distinction between CDS and proper insurance policies.

• First, the obvious difference between a CDS and a proper insurance policy, such as flood or fire insurance on a home, has to do with the fact that a CDS is synthetic while a home is “funded”. What that means is that I can write as much CDS as my heart desires (the outstanding principal of a bond has no bearing on how much CDS can be traded) while you can write only a single insurance policy on a home. This automatically limits the amount of exposure insurance companies can take on (ignoring further the reserves they need to hold against this policy).

• Second, lets’s say you are a proud owner of a fire insurance policy on your home and one day you spot a man having a cigarette 30 feet from your home. Will you call your insurance company and demand collateral given the increased risk of fire? Probably not, but this is what happened in the case of CDS (yes I am simplifying, of course)

On the collateral side, clearly it was a mistake to let AIG and the monolines not post collateral. Establishing a clearinghouse will make sure this won’t happen in the future. However, apart from these two cases, margin requirements on CDS do exist and are followed rigorously. Hedge funds do post collateral to dealers when they trade CDS. It’s true that some hedge funds have to post minimal amounts, however those funds open up their books to dealers. In fact, in the case of Lehman, ISDA commented that 2/3 of the CDS exposure was collateralized.

But the CDS market is so opaque and unregulated

Tell that to the DTCC that have been documenting gross and net notional amounts of the vast majority of CDS trades since 2006.

DTCC

But allowing AIG to fail would have caused AIG’s counterparties to fail as well, given the $100bn+ payouts

This is actually more subtle. First, the payments to AIG’s counterparties don’t accurately represent each party’s exposure to AIG. For example, if the government paid $13bn to Goldman, it does not mean that Goldman would have lost $13bn had this not happened. This is due to the following:

• Some protection written by AIG is likely to have been collateralized. In fact, this is what Goldman has claimed.

• Banks likely bought protection on AIG to protect themselves in case of AIG failure. A bankruptcy by AIG will have allowed the banks to monetize this protection.

The problem facing the banks had AIG failed has less to do with their $ exposure to AIG and more with their position exposure to AIG. For example, let’s say I buy $100mm of protection from AIG and then I buy protection on AIG to hedge against the case of AIG’s bankruptcy. Let’s say AIG does in fact go bust. In the ideal scenario the collateral plus the AIG hedges offset exactly my CDS MTM exposure against AIG, then the banks don’t actually lose money. However, the problem is that they are now long risk $100mm of protection (because their $100mm short risk position against AIG is now gone). What happens then is the market realizes that a dozen banks have massive long risk positions in much of the same trades that they will all now try to hedge at the same time. Spreads blow up and the banks lose.

Fianlly, I am ignoring such comments as Bear Stearns and Lehman collapsed because of CDS or that CDS can be used to drive companies into bankruptcy both of which probably don’t deserve discussing.

The recent widening in United States Credit Default Swap levels has gotten a lot of attention once it cleared the magic 100bps level intra-day.

As with any CDS-related news, you will get heated commentary in the blogosphere with a large perception of folks simply calling for all CDS trading to be banned. The general consensus appears to be “don’t the buyers of CDS realize that in the event of default by US, these contracts are not likely to be honored anyway?” This is Krugman’s line. Taleb chimes in with “It would be like buying insurance on the Titanic from someone on the Titanic”.

As with any heated commentary there’s bound to be a lot of misunderstanding of what this recent widening actually means and where it comes from. I’ll try to tackle this issue point by point below. For those of us with ADD (myself included) here’s a brief summary:

• Traders don’t buy CDS because they think the name will default; they buy CDS because they think the spread will widen – I make this point in my AIG post. It follows that extrapolating any default information from wider CDS spreads can be misleading
• An apples-to-apples comparison of US CDS spreads suggests that $-denominated US CDS (the standard  contract that is quoted in the news is the €-denominated one) should be trading at half the level it is now, perhaps making the recent news a lot less exciting
• The standard CDS contract is sufficiently complex so that the end-game buyers of CDS can be betting on something much more innocuous than a “default” such as a restructuring of privately negotiated tiny-size debt issuance
• Sovereign CDS (US included) has actually lagged both rising financial as well as systemic risk and has only now caught up, making the recent move largely expected

CDS is not a “default” trade – it is a “spread” trade
The most important point to be made here, the same one I make in my AIG post, is that, one shouldn’t look at CDS as a “default” trade. Though their pricing is clearly driven by the likelihood of default and the payout upon default, I can tell you that 99% of people buying CDS do not believe that the entity upon which they are buying protection will actually default. In this, they are similar to investors in stocks. People buy and sell stocks because they think the stock in question will increase or decrease in price. Same goes for CDS.

I think the confusion largely stems from people viewing CDS akin to insurance. Though this is an easy analogy to make, it is, in fact, wrong. What motivates people when they buy fire insurance is that, in the unlikely case their house is consumed by a fire, they will get reimbursed. This is not what drives the CDS market.

There are two key differences between CDS and the insurance analogy:

1. I don’t need to have a position in the entity’s bonds or loans in order to trade CDS on the same entity (while I do need to own the house I buy fire insurance on)
2. As I mention above the vast majority of traders don’t trade CDS because of a view on default – they trade CDS because of their view on the level of CDS spreads expecting to lock in a MTM profit on the trade. Though you can probably save yourself some premium on fire insurance by installing sprinklers it’s clearly not as easy to do nor is it the primary motivation for fire insurance in the first place

Sovereign CDS is not a “fundamental” trade
I think one thing we can safely dismiss as the driver behind the widening of US CDS spreads, or in fact any sovereign spreads, in the market is any kind of fundamental view of where these spreads should be. The difficulty behind trading CDS on a fundamental default probability basis has to do with the fact that in order to put a number on an absolute default probability you need to have a firm view on: a) default likelihood, b) recovery upon default, c) devaluation of the local currency, to the extent that CDS you are trading is denominated in local currency.

Going through these in order

• It is actually difficult to have a firm view on the absolute default probability of any sovereign, particular, the United States. The fact is that developed sovereign defaults are relatively rare (outside of Spain’s relatively orderly 6 defaults within 100 years starting in the 16th century). As far as United States, my best guess is that we would need to go back to the Civil War to find a proper case of a “default”, though even here you would have to stretch. This was when the Confederacy issued cotton-backed bonds to finance the war against the North. Once the South lost New Orleans (making it impossible for South’s creditors to take physical delivery of cotton) and began to run out of cash, it became clear that it was only a matter of time before the Confederacy defaulted. By the end of the war the Confederacy’s “greybacks” were worth 1 cent on the dollar. The North refused to honor the Confederacy’s debts and the rest is history. In the 20th century developed sovereign defaults are relatively rare, especially after the World War II.

European defaults/restructurings in the 20th century (Rogoff)

• Getting a guage on expected recovery by a sovereign is not any easier. These range from the teens in Russia and Ivory Coast to 69% in Ukraine.

• Though much of protection traded on sovereigns is in a currency other than the local currency (i.e. Brazil CDS is traded in USD not in BRL), for local currency trades one has to be aware of the likely devaluation of the local currency in case of default. Those of us old enough to remember will recall the Argy peso going from 1 to over 3 in its peg to the dollar. I touch upon this in the Quanto CDS but suffice it to say that buying protection on Germany in EUR rather than USD means that €CDS levels should trade around half of $CDS levels.

The Quanto CDS
Though people like to focus on the round 100bps number, what’s mising from this is the fact that US CDS is traded in euros and that in order to do a proper apples-to-apples comparison to US-traded corporates you would need to first translate the EUR spread to a USD equivalent spread. This translation is largely a function of how much the local currency will devalued in the case of default (ignoring the small impact of rate, fx and credit volatilities and correlations). So, if you think that the dollar will weaken by 50% relative to the Euro in the case of a US default then the fair USD-denominated US CDS spread should trade around 50bps.

Do sovereign CDS trade in currencies other than the standard contract currency? In fact they do and the biggest market is in Latin American CDS denominated in local currency. If you think CDS is an “obscure” market, then this is the ultra-obscure one. It is largely driven by sovereign issuance of US-denominated debt that they swap to their local currency (in order to remove the stain of “original sin” ie non-local-ccy issuance). Normally, they would just do a simple USD/local-ccy interest rate swap. However, the trick is to do a clean asset swap instead which is simply an interest rate swap that is credit-linked to themselves which can save the country upwards of 100bps on the swap. Corporates in Europe and Latin America tend to do this “self-reference” trick as well – though it is illegal in the US.

For Latin American CDS, this local currency discount can be anything from 25-60% on 5y CDS (it varies depending on the tenor and tends to be downward sloping).

The “non-default” default
The word “default” has been thrown around a little too easily lately with respect to CDS contracts. The concept of default is, generally speaking, a very loaded one that brings to mind long bread lines, a crippled banking system and runaway inflation. In the context of CDS, the concept of “default” is a very specific one. For this reason, CDS language talks about a “credit event” rather than a “default” and can include such actions as restructuring of debt, repudiation of debt, moratorium and accleration. In summary, the following issues need to be considered in the context of Sovereign CDS.

• The nature of the “credit event”. For Western Europen sovereigns, for instance, these include a) Failure to Pay, b) Repudiation/Moratorium, c) Restructuring. Latin American sovereigns add to this list Obligation Acceleration which was a near possibility when Hugo Chavez declared his country’s pullout from the IMF. The point here is that something like a restructuring of debt can be much more benign than an outright default (i.e. a failure to pay). So, a CDS can often price in a less dire scenario than the likelihood of “default”.
• Generally, anything counting as “Borrowed Money” can trigger a CDS credit event. This can often be a small privately negotiated loan rather than a large bond or loan trading in the market.

The Beta Issue
If you ask a Sovereign CDS trader why his names are wider today his likely response is “The index is blowing up, dude. Now do you have anything to do?” The simplest answer is that sovereign CDS is wider because everything is wider. And the simplest answer is often the right one.

Backstopping Financials
The move wider in sovereign CDS can be attributed to the expected covergence between sovereign and bank CDS spreads on the back of countries backstopping their financial systems. Countries have either bailed out certain institutions directly (Lloyd’s, RBS, ING, etc.) or have guaranteed bank deposits (Ireland, Germany,e tc.). While sovereign spreads have initially lagged the spreads of their financial systems, once it became clear that the sovereign was willing to underwrite the tail risk of their banks, it made sense for their spreads to converge. And if financial spreads refused to come down to the level of the sovereign, then sovereign levels would rise to the level of financial spreads. This was likely driven by two things: a) relative value trades of selling bank CDS and buying sovereign CDS betting on the convergence, b) continued buying of bank CDS as a hedge against bank paper. This led to sovereign CDS widening to the level of bank CDS rather than the other way around.

The Systemic Hedge
One way to understand the widening in sovereign spreads is by tieing sovereign risk to some other risk in the market that should be driven by the same views or needs. The typical buyer of sovereign CDS, apart from the marginal trader punting on Austrian eastern european exposure of the inability of  Iceland to convince the world they’ve got things under control are the Investment Bank credit portfolio groups. These departments generally manage hundreds of billions of loan and derivative exposure across the bank. Their mandate is to protect the bank from an increase in non-performing loans. Normally, the counterparties to the loans do not trade in the market (either in CDS or stock) or are not liquid enough for the groups to go out and hedge  in these assets. So, what they normally end up doing is buying systemic risk hedges in large size with the expectation that in the scenario a large portion of the bank’s loans goes bust, the world will be in such a state that their systemic hedges will offset the deterioration in the loan book. Though out-of-the-money S&P puts figure prominently in their hedges, in the world of credit we can look at a) super-senior spreads, b) financials spreads, c) sovereign spreads.

Assuming 40% recovery, the CDX super-senior tranche (30-100%) will be impaired after 40% of the CDX portfolio. Although it’s clearly difficult to envision the state of the world in this scenario, we can safely say the sovereign would be under pressure.

Why is U.S. Credit Risk News?
It is interesting that US credit risk is showing up on people’s radar at the moment when the “obscure” product like CDS is signaling it rather than the plain-vanilla Interest Rate Swap which trades in many multiples of volumes.

Sometime in early 2009, 30y interest rate swap yields fell below treasury yields. This price action suggested that the market viewed 30y Bank (AA) risk as safer than Treasury (AAA) risk. However, given the dire state of the Banks, this was clearly not the driver of the yield moves. What happened was that the exotics desks of the banks sold a huge amount of 2s/30s non-inversion notes to private bank investors that paid a high coupon as long as 30y swaps stayed above 2y swaps. The initial hedges done by these desks was to pay 30y swaps and receive 2y swaps. By the end of the year, rates had collapsed with 30y swaps falling more than 2y since the front end did not have as much room to rally. This meant the 2s/30s curve flattened massively causing the banks to partially unwind the hedges. In a period of poor liquidity every rates exotics desk was hitting 30y bids in size leading 30y swaps to rally beyond treasury yields.

So, though often painted as a credit risk issue, this episode was really a liquidity/technical problem.

Bring on the Technicals
Here, I briefly describe what, in addition to the above issues, could be the technical drivers of wider sovereign CDS, and US CDS in particular:

• Credit-Linked Notes Unwinds. As we all know retail investors are the best negative gamma traders. They buy high and sell low. It is not impossible that there were investors who were looking to add a few basis points to their “risk-free” trade by adding US CDS risk. It is also possible that as the crisis deepened they grew less comfortable with the risks in the trade and unwound them, suggesting that the origination desks needed to buy back the US CDS protection they initially sold, pushing CDS wider
• Liquidity in US CDS is not fantastic judging by two things: a) US dealers don’t trade it and b) the bid/offer spreads is 10bps or around 12% of the CDS spread. By comparison bid/offer spread in the CDX index (most liquid product in credit) is less than 1%. When you factor in these issues with the fact that in the current environment there are likely to be more buyers than sellers, you will see the CDS spreads widen to accommodate that

So, in summary, what do I make of US CDS widening? Well, not much apart from making it another cocktail conversation topic. Let’s revisit this issue once investors start discounting all their treasury holding by the US CDS spread… starting with China and their $1.7trn portfolio. Now that would give us something to talk about!

WSJ

Fast forward to early 2009 and Boaz Weinstein, the former star trader and co-head of credit trading at Deutsche Bank is down $1bn, Ken Griffin of Citadel is down 50% and John Thain’s Merril is said to be down $10bn+. Most of these horrific losses are due to a single strategy… the scary negative basis trade.

Bloomberg has written about it here

And there has even been a book published on this strategy… how many trades can say that!

What is a negative basis trade?
In a nutshell, if a bond is trading cheaper than the CDS you buy the bond and buy CDS protection to lock in “risk-free” carry. In other words, say there is a 5y bond in the market trading at 500bps spread while a 5y CDS is trading at 480bps. Putting on this trade gives you a carry of 20bps, which if you lever it up 25x + provides you with carry of 5%+.

The reason this is called “risk-free” is because you don’t care (caveats coming) if the name defaults since what you lose on the bond you make back from the short risk position in the CDS.

Why is it called negative basis?

This is called a “negative basis” trade because “basis” is the difference between the CDS spread and bond spread (CDS spread – Bond spread). This basis is negative when the CDS spread is lower than the bond spread. This means the bond is cheaper than the CDS and if you buy the bond and buy CDS, you are left with a positive annuity. Doing the positive basis trade (shorting the Bond and selling CDS protection) is tougher as it requires shorting the bond which runs into messy repo issues etc.

History of Bond-CDS Basis for High Grade Bonds

Below is a chart of the behavior of the Bond-CDS basis for high grade bonds in the last two years:

What typically drives the basis

The basis for most names has tended to be negative largely owing to the following factors most of which will tend to cause the basis to go negative:

• CDO issuance: the so-called synthetic bid or large issuance of synthetic CDOs meant that there was huge demand by client to sell CDS protection. This meant that CDS spreads tightened relative to Bonds
• Bond issuance: heavy issuance puts more pressure on bond spreads
• Leveraged accounts: (hedge funds etc.) who borrow above Libor will find it more attractive to sell CDS protection rather than buy bonds.
• CDO unwinds: CDO unwinds will push CDS spreads wider i.e. the basis more positive. During this last cycle I have seen many more restructurings than unwinds which allow investors to maintain the rating on the structure if the original one has been downgraded and keeps them from locking in MTM losses. This is one of the reasons why the negative basis has persisted
• Repo: Putting on the positive basis trade is more difficult as you run into issues for hard-to-borrow bonds. You also have repo roll risk as getting term repo is not always possible.

Calculating the basis (wonkish)

The definition of the Bond-CDS basis is clear enough: CDS spread less the Bond spread. What the CDS spread is obvious – it’s easily observable in the market. What is not obvious is which Bond spread to use as we can potentially choose from the smorgasbord of I-spread, Z-spread, Asset swap spread, and others.

Remember, our goal is to find a bond spread that best matches the CDS spread so we can do an apples-to-apples comparison. As I briefly discuss below these measures are not up to the task:

• Z-spread: is effectively how much we need to parallel shift the risk-free curve so that the present value of the bond’s cashflows matches its traded price. This measure does not take into account the term structure of default probabilities nor the expected recovery upon default.
• Asset swap spread: is the spread over Libor received on an asset swap package (which transforms a fixed-rate bond into a floating rate note). Much like the Z-spread above this measure does not take into account the term structure of default probabilities nor the expected recovery upon default.

What is needed for a “perfect” bond spread measure that allows an apples-to-apples comparison to CDS are the following characteristics:

• It should follow the same convention: ie day-count (A/360 as in CDS, not B/360), accrued paid upon default (yes in CDS, no for Bonds), etc.
• It should take into account the term structure of default probabilities
• It should take into account the expected recovery upon default

While it’s not completely trivial to calculate this spread, it’s doable. And conveniently enough several broker-dealers are nice enough to do this for you.

Things to think about in a basis trade

Running a negative basis position is not trivial and can take up some cranial capacity. Briefly, here’s what you need to be aware of:

• Maturity mismatch – normally you cannot buy CDS (which trades at IMM maturities) to the same exact date as the bond maturity
• If CDS trades upfront the math gets a little messy
• You have interest rate exposure and potentially FX exposure
• Restructuring and deliverability issues: upon default your bond may not be deliverable into the CDS contract

What went wrong

• Lower leverage: a year ago if you were a hot-shot hedge fund you were able to lever a negative basis trade 25x +. Then your prime broker gave you a call and told you that the new leverage is 5x… so come up with the cash or cut the position. Most funds cut the position driving the basis against them (and other funds)
• Funding cost: the financing cost in the good old days was Libor flat and now… Libor + 100 making the trade much less economical
• Margin on CDS: Buyers of CDS protection never had to post margin before - why would you you’re buying insurance right? Well, wrong! The MTM can still go against you so pony up that margin please…
• Redemptions: If you’re a hedge fund facing redemptions you have tended to cut your most liquid cash positions…which included high grade bonds.

So, the moral of this story is probably if your strategy depends on high leverage and abundant liquidity think twice before piling into it. Though my view is that the basis should recover from current levels, everything depends on the stress in market confidence and broker-dealer health.