Restructuring and bankruptcy

The Economics of Contempt has a nice historical summary:

the first major restructuring that I can remember being significantly hindered by CDS was Marconi, and that was back in 2001-2002. Marconi was negotiating a restructuring with a bank syndicate, but for a long time certain syndicate participants (cough, UBS, cough cough) refused to agree to any restructuring that didn't constitute a "credit event" under the 1999 ISDA Credit Derivatives Definitions. The holdout banks had purchased CDS on Marconi to hedge their exposure, and if they were going to agree to a pretty drastic restructuring, they wanted to make sure they got the benefit of their hedges. After more than a year of restructuring negotiations, the banks agreed to a debt-for-equity swap that qualified as a credit event under most of the CDS contracts, but also pretty much wiped out shareholders.

Mirant Corp.'s 2003 bankruptcy was also largely a result of CDS. Several creditors had purchased CDS protection on Mirant, and one major creditor in particular, which rhymes with Pitigroup, was relatively open about the fact that it didn't agree to a restructuring because it needed a bankruptcy filing to trigger its CDS contracts referencing Mirant. The bank that rhymes with Pitigroup's refusal to agree to a restructuring (which came at the last minute and was a big surprise, if I remember correctly) effectively torpedoed any chance Mirant had of avoiding bankruptcy.

I'm not personally familiar with Mirant, but the Marconi example is certainly a good one. There is a definitely a good case that rights to sit at the creditors table should sit with the risk holder, not the bond holder.

Honour

Martin Wolf, quoting Mervyn King, writes in the FT:

'My word is my bond' are old words: 'My word is my CDO-squared' will never catch on.

He's right, but it is a shame. After all, we need a convenient shorthand for 'My word is badly structured and likely to lead you into trouble'.

Post of the day

How can you not like something that begins 'Emerging Markets and derivatives are like alcohol and barbiturates: each on its own has attractions but create a recipe for choking on one's own vomit when combined'? It's about nefarious doings in the CDS market on the Kazakh bank BTA, and it is here.

Before the bust: AIG's early collateral postings

AIG's collateral postings after the rescue are well known - essentially the firm was saved so that it could continue to fulfil its obligations to the banking system, notably under the CDS it had written. AIG before the fall has received less attention. But now Bloomberg has done some digging, and the story of the collateral calls that brought AIG down is emerging.

Goldman Sachs Group Inc. and Societe Generale SA extracted about $11.4 billion from American International Group Inc. before the insurer’s collapse as the firms demanded to hold cash against losses on mortgage-linked securities, ... “It was precisely that drain of liquidity to Goldman and SocGen that put AIG in a position of illiquidity and ultimately threw them into the government’s arms,” said Charles Calomiris, a finance professor.

Including collateral from before and after the rescue and payments made by Maiden Lane III, a vehicle created by the Fed to retire the swaps, Goldman Sachs received about $14 billion from AIG, Societe Generale got $16.5 billion, and Deutsche Bank AG received $8.5 billion.

The Amherst Trade

This is a geeky post about the CDS market.

The newswires have been buzzing recently with news of a 'daring' CDS trade by Amherst Holdings. I didn't comment on it at first as I didn't understand the trade from the initial news items, but I now think it is possible to work out what's going on.

Let's start with the bonds this trade refers to. They are subprime MBS. Like most MBS, these are amortising, prepayable bonds. The fact that these are amortising bonds means that the face value of the security is irrelevant: what counts is the principal balance at the time the trade was done. [Quite a lot of the stories were confused about this point, so it is worth pointing out.]

So, let's say we have some bonds with $100M left to repay.

The next bit that is tricky is the nature of the CDS protection sold. Everyone agrees Amherst sold protection and some banks bought it. But what protection exactly? It is most common for CDS on MBS to be pay as you go, meaning that the protection sellers compensates the protection buyer for principal deficiencies as and when they occur. There is no event of default as such, unlike corporate CDS. [To be strictly honest, there may be an event of default as well, such as bankruptcy of the issuing SPV, but that is irrelevant for our purposes.]

Let's suppose then that Amherst sold pay as you go protection on $100M of bonds.

Since the bonds were thought likely to repay little to nothing of the outstanding principal balance, the banks paid Amherst, say, $80M up front for their protection. [There may have been an ongoing coupon as well, but we'll ignore that.]

Amherst then paid the servicer to buy out the underlying mortgages and pay off the bonds. Thus the bond holders got their $100M. The servicer could do this because the bonds had a 10% clean up call, meaning that if more than 90% of the face had amortised, they could repay the remaining principal balance at any time. [So to keep with the example, the face amount was more than $1B.] 10% cleanup calls are common in ABS, and they are what makes Amherst's trade work*.

Now, here's the confusing part. Who won and who lost?

First the banks. If they had held the bonds, then they would be about flat. $80M for CDS protection paid out, but $100M paid back is a $20M profit, from which subtract the (few cents) cost of the bonds. So the only way the banks could have lost massively on the trade, as reported, would have been if they had been net short the bonds. That is, they did not own the bonds, and bought protection, betting that total losses would be more than $80M. The losers, then, were parties who did not own the bonds and who did not realise the significance of the cleanup call to their short.

[The WSJ story suggests that JP Morgan lost money but that RBS and BofA didn't. This would be consistent with JP being net short, while RBS and BofA had a negative basis trade on, i.e. owned the bonds and bought CDS on them. The presence of net shorts is also consistent with the WSJ's suggestion that more protection had been traded than the notional of bonds outstanding.]

Next Amherst. They had the $80M of CDS premium. But how much did they have to pay to get the bonds repaid? Clearly a logical answer would be about $80M. Therefore the only way that Amherst could have made money would have been if they had sold more protection than there were bonds - $200M say rather than $100M. Say they sold $50M to JPM, $50M to RBS and $50M to BofA. Then they would have had to pay $80M, roughly, to buy back the mortgages behind the RBS and BofA CDS, but the JPM CDS was not backed by any bonds and so the $40M premium from JPM would be straight profit.

In other words, the only way Amherst could have made a lot of money on this trade would have been if it sold more protection than there were bonds. The only way that the banks could have lost money would have been if they bought more protection than there were bonds. In a situation like this someone was always going to be squeezed. It's just that this time, that party wasn't an investment bank.

The lesson of this amusing little situation? Nothing more than read the small print. The buyers of protection -- and in particular naked shorts -- should have understood that arbitrary action by servicers is possible, and that in particular the 10% clean up call could be exercised. This is a much bigger risk late in the amortisation profile of a bond than early, but it is there for most ABS. Caveat emptor.

* Contrary to what Willem Buiter writes in his blog, if you own 100% of a bond, you cannot necessarily control whether it defaults or not. A default on a public security is a default, regardless of who is affected.

Another mistake Buiter makes is assuming that centralising CDS trading would not have helped in this situation. It would certainly have helped the banks to avoid their losses, in that the size of Amherst's long vs. the cash would have been obvious thanks to trade reporting. Personally I don't particularly feel the need to help the CDS trading desks of investment banks, mind you.

CDO waterfall errors

CDOs are complicated. In particular, many of them have waterfall structures that include diversion tests - if x then tranche y gets some money, otherwise it goes to tranche z. Unsurprisingly, trustees sometimes get these tests wrong. Expected loss has found an example: I do encourage you to read it if you have an interest in either structured finance or operational risk.

Credit protection sanity

I have made a bit of a sideline in highlighting some of the less helpful comments on the CDS market. It pains me to include a man I generally respect, Paul Krugman, in the list of people who have got the wrong end of the stick.

He first quotes marketwatch:

The spreads on credit-default swaps for U.S. government debt jumped to 97 basis points Tuesday, nearly seven times higher than a year ago and 60% higher than the end of last year, to a level roughly in line with those of France, according to data supplied by Markit.

Then he opines:

Has the risk of a US government default risen? Probably. Nonetheless, the people buying these contracts are crazy. A world in which the US government defaults would be a world in chaos; how likely is it that these contracts would be honored?

The answer is that it does not matter (much). Most CDS trading is about views on the spread, not views on default. People buy CDS on the US government because they think the spread will widen and they can close out at a profit, not because they think that default is likely. Therefore the CDS market often tells you rather little about default: what it tells you about is market participants expectations of spread movements. CDS spreads go out when there are more buyers of protection than sellers. That is the only reason spreads move. Any connection between CDS spread movements and expectations of default is a modeling assumption, and one that is particularly dubious at the moment.

Fascinating

Corporates trading through their sovereign in the CDS markets, from Zero Hedge:I don't like the obvious trade (long protection on the corporate, short on the sovereign) where the corporate can easily raise liquidity offshore, but there are still some opportunities here.

Friday in Fantasy Land with George

I should buck up, I know. The current furore over credit derivatives may well come to nothing. But when I read something like Soros' article in the FT today, my heart sinks. Let's play spot the fallacy.

The second step is to understand credit default swaps and to recognise that the CDS market offers a convenient way of shorting bonds. In that market the asymmetry in risk/reward works in the opposite way to stocks. Going short on bonds by buying a CDS contract carries limited risk but unlimited profit potential; by contrast, selling credit default swaps offers limited profits but practically unlimited risks.

The asymmetry encourages speculating on the short side, which in turn exerts a downward pressure on the underlying bonds.

So how can you explain the growth of CPDCs, who are long risk only CDS investors, or for that matter the long risk portfolios at the monolines and AIG? In reality the fact that CDS allows unfunded risk taking trumps the ability to go short in stimulating demand.

Note too that there can be no asymmetry in the market: you can't sell CDS protection unless someone buys it and vice versa.

...US and UK government bonds... actual price[s are] much higher than that implied by CDS. These asymmetries are difficult to reconcile with the efficient market hypothesis, the notion that securities prices accurately reflect all known information.

No they are not. (I don't believe in the strong form of the efficient market hypothesis either, but this is not evidence for or against it.) CDS spreads include counterparty risk effects and liquidity risk on the possible future margin. Bond spreads include funding premiums. They are literally incommensurate since to have hedged positions (bond plus CDS) you need to be able to fund the bond to term at a fixed cost and to have no counterparty risk on the CDS provider.

I have an unworthy, low conjecture. It is that the age of a commentator is directly related to their hostility to CDS. The older someone is, the later they are likely to have come to the CDS market. And the less likely they are to understand it.

Negative basis hell

There have been a few posts around recently about losses on negative basis trades. These were apparently (one of) the source(s) of Merrill's recent loss. But I had not seen a convincing explanation until I came across this post from Zero Hedge. This picture is particularly eloquent:

At the time of the Lehman bankruptcy, it shows the spread had blown out over a thousand basis points. Now, given there is little liquidity in this stuff, Merrill is unlikely to have taken most of the trades off. So if this is the main cause of the losses, they will come back as the basis gets back to normal.

More awful journalism on credit derivatives

The ignorant journalist's favourite whipping boy, the CDS market, gets another undeserved beating today, this time from Alan Kohler in the Business Spectator. Managing a considerable density of mis-information, Mr. Kohler is rather histrionic:

As the world slips into recession, it is also on the brink of a synthetic CDO cataclysm

Would that be a cataclysm like the Lehman CDS one, or like the Fannie and Freddie one? How would sir like his non-event today?

The triggering of default on the trillions of dollars worth of synthetic CDOs that were sold before 2007 could be a disaster that tips the world from recession into depression. Nobody knows, but it won’t be a small event.

Or just perhaps it will be totally orderly just like all the other CDS settlements so far.

CDOs were invented by Michael Milken’s Drexel Burnham Lambert in the late 1980s

No, CMOs were invented by the Freddie Mac in 1983. The first CBOs were based on that structure.

About a decade later, a team working within JP Morgan Chase invented credit default swaps, which are contractual bets between two parties about whether a third party will default on its debt. In 2000 these were made legal

Again no. They were legal when they were first traded, in the early 1990s. (U.S. law, it may surprise Mr. Kohler to learn, is not the only relevant one.) At this point I have to confess I gave up. Anyone who makes that many basic errors in the first few paragraphs while using over-blown and emotive language does not deserve to be read.

AIG and default correlation mis-estimation

Felix Salmon has a nice piece on AIG FP's strategy and why it went so badly wrong.

When AIG wrote protection on CDOs and the like, it got insurance premiums in return, and considered those premiums to essentially be free money, since (according to AIG's own models, and those of the ratings agencies) the chances of those CDOs defaulting were essentially zero.

...AIG's biggest mistake was in failing to realize that this business couldn't scale in the way that most insurance does scale. Most insurance does scale: if you insure a house against fire, for instance, it's easy to lose much more money than was paid in insurance premiums. But if you insure houses across the country against fire, you'd need a nationwide conflagration in order to lose lots of money.

... The reason AIG's models said the CDOs couldn't suffer any losses was that house prices don't fall in all areas of the country simultaneously. Since AIG was only insuring the last-loss CDO tranches, investors with lower-rated tranches took the risk that prices in Florida, or Arizona, or California might fall. AIG would only lose money if prices fell in all those states at once -- which is, of course, exactly what happened.

In other words, AIG's models assumed default correlation would be low, and that there was a good measure of diversification benefit between the different CDOs it had written protection on. In reality once house prices turned down there was very little diversification, default correlations leaped up, and the mark to market on many of AIG's contracts turned against them, necessitating the collateral postings that brought the insurer into the welcoming arms of the FED.

That AIG bailout again

AIG wrote CDS protection to a bunch of banks. Spreads have blown out and AIG is not AAA, so they have to post collateral. They couldn't, so the FED lent it to them at Libor + 850. That was last month's story.

This month's story is a new bailout. Let's see if we can follow the story in the WSJ:

Many banks that previously bought protection from the insurer on securities backed by now-troubled mortgage assets stand to recoup the bulk of their investments under a plan by AIG and the Federal Reserve Bank of New York to buy around $70 billion of those securities via a new company. These securities are collateralized debt obligations backed by subprime-mortgage bonds, commercial-mortgage loans and other assets.

...

The banks also will sell the CDOs to the new facility at market prices averaging 50 cents on the dollar. The banks that participate will be compensated for the securities' par value in exchange for allowing AIG to unwind the credit-default swaps it wrote.

So the Journal is suggesting that the new facility will buy the securities and cancel the CDS AIG has written. Note that that does not just get the banks off AIG counterparty risk: it also frees up funding for the underlying assets. If eventual recoveries are greater than market prices predict, then the new entity at least gets to keep the difference. Still, it does seem a strange way to proceed. Even if you thought that it was important to ensure that the protection buyers did not lose - clearly a key feature of the bailout - why would you buy the underlyings rather than simply loan AIG money to meet collateral calls and recapitalise them if necessary?

How many ABX CDS are there?

Alea had a misleading note on the DTCC reporting of credit derivatives. I usually like Alea, but this post, suggesting there were only single digit numbers of contracts on many of the ABX subindices, gave the wrong impression. Turning to the DTCC itself, we find net notionals in most of the subindices in the single digit billions, with high hundreds or low thousands of contracts.

What are the dynamics of risky bond prices?

The Basel Committee's two papers on incremental risk in the trading book (incremental to that captured by VAR, that is) - here and here - led me to muse on what the real dynamics of risky bond prices are.

Firstly clearly there is an interest rate risk component. Let's ignore that as it is the best understood.

Second there is jump to default risk. The phrase itself is slightly misleading in that bond prices often fall a long way in the period before default, and indeed recoveries are sometimes higher than pre-default bond prices would suggest. Skip to default might a better term. Still, the idea that there is a jump process which can cause non-continuous changes in risky bond prices is reasonable.

Then there are 'everyday' movements in credit spreads. Now, here's the six hundred and forty billion dollar question (OK, OK, not the size of the corporate bond market I know) - if you take out the jumps, is what you are left with even vaguely normal? My guess is that it isn't, and that autocorrelation is significant even after jumps have been taken out. The hard part is that you need a lot of credit spread data to look at this kind of thing, and it isn't easy to come by. CDS data won't do in this instance simply because single name CDS have only been liquid for ten years or so, and you'd at least want data going back well before the '98 LTCM/Russian crisis. I'll get around to this sometime soon...

Spread dynamics are the flipside to my earlier post on what CDS spreads mean: that was about what causes the spread to move; this is about how you can model those movements.

What do CDS spreads mean?

Naked capitalism suggests that some observers are perplexed by widening CDS spreads on US government debt.

How, they ask, could a private sector contract against default be expected to pay out in the case of a US government default – which would be the equivalent of a nuclear explosion in the financial markets?

The ten year Euro-settled spread has gone out considerably:All that is going on, of course, is that people are buying the contract because they think the spread will go out further. That will happen if more people buy it. More buyers than sellers equals rising prices. The actual default of the US has little to do with it for the buyers: they care about spread movements, and making money from them. Sellers are presumably happy to take what they view as an immaterial risk.

Update. Bloomberg has an article which confirms the idea that a lot of CDS trading is driven by speculation on the spread rather than insurance against default. They report that the most active contracts recently include those on Italy and Spain. Dubious though both countries' finances may be, default from either is vastly unlikely - but spread widening is rather likely.

Credit Derivatives Unfairly Scapegoated

From Reuters:

Credit strategists at ING said on Thursday that credit derivatives were being unfairly blamed by politicians and commentators for the near meltdown of financial markets..."The CDS (credit default swaps) market is being used as a scapegoat for political and economic goals," ING credit strategist Jeroen van den Broek wrote in a note to investors.

Quite right too, and I have been saying so for months. This particular goat (sheep, whatever) is pretty safe.

Update. As per my earlier post, Reuters reports that the Lehman settlement is a non-event. There a nicely written elaboration from Felix Salmon on portfolio.com here, and a broader FT alphaville defense of derivatives here.

Credit Derivatives Still Not Spawn of the Devil

Let me offer an analysis of much of the media coverage of credit derivatives recently.

"Credit derivatives traders eat babies. Now XYZ event will tip the financial system further into chaos thanks to these dastardly instruments..."

(XYZ = Fannie/Freddie credit event, Lehman credit event, ...)

"On the eve of the auction for XYZ we spotlight the satanic unregulated* credit derivatives market..."

[The event happens. Settlement is orderly.]

"XYZ event was OK, we suppose, but the NEXT event will truly cause the end of the world thanks to the inexorable evil of credit derivatives."

The DTCC has an antidote. I doubt many bloggers or mainstream journalists will read it.

* Where by 'unregulated' we mean of course 'unregulated apart from the regulation'.

If the US defaults, people will still want soup

From the Economist a couple of CDS spreads:

Campbells Soup 17 basis points

United States of America 19.8 basis points

(And Morgan Stanley is over 2000, if you can get it.)

Fannie & Freddie sub recovery bigger than senior

From Reuters, initial indications for today's auction:

	Fannie	Freddie
Senior	92.4%	93.75%
Sub	92.65%	93.8%

Remember, kids, while senior unsecured recovery can't be lower than sub in the actual bankruptcy hearings, it surely can in a CDS auction.

Update. Final results in:

	Fannie	Freddie
Senior	91.51%	94%
Sub	99.9%	98%

Alea suggests that the recovery mismatch is due to a cheapest to deliver effect as the zeros are deliverable into the senior swaps (list of deliverables here). Personally I think it is partly because there is a lot more liquidity on the sub than the senior. In particular naked shorts played with the sub (partly to hedge nationalisation risk). A lot of those are short covering now, whereas the senior was typically a credit risk management trade rather than speculation.