Does Lynton Jones even know what a hybrid ABS CDO is?

There is a mind-bogglingly deceptive report issued by the City of London blaming the crisis on "over leveraged" ABS CDOs and claiming that this is evidence that credit default swaps did not lie at the heart of the crisis. This is a ridiculous point of view. Credit default swaps were the tool used by ABS CDOs to become over leveraged. While it is possible to argue that certain categories of credit default swaps are not subject to abuse (e.g. that this was a problem of index CDS and that single-name CDS are inherently safer products) or that some targeted forms of regulation are sufficient to address the problem, it is dumbfounding for someone to claim that the crisis was caused by CDOs and not by the credit default swaps they used to become overleveraged.

The confusion created by this misleading report has already led the FT to publish a commentary by someone who clearly does not understand how ABS CDOs were structured because he claims based on the City of London report: "Let’s start with the simple proposition, which seems to be almost universally accepted, that the OTC markets in credit derivatives were one of the main causes of the crisis. The City of London report utterly refutes this claim."

Why are assets toxic?

An explanation for those who can’t understand how so many assets backed by mortgages can be worth nothing at all

From 2005 through the first half of 2007 approximately $2.8 trillion in mortgage backed securities were issued by the private sector. These securities were backed by $1.1 trillion in sub-prime loans, $0.9 trillion in Alt-A loans and $0.6 trillion in prime jumbo loans (and $0.2 trillion in “other” loans). Data in intro here.

I estimate that these $2 trillion in non-prime loans and $0.6 trillion in prime jumbo loans back $4.1 trillion in AAA mortgage related securities. How is it possible to have more AAA assets than underlying loans? Through the magic of credit default swaps.

I should be clear, however: I am including in my $4.1 trillion figure $1.75 trillion of unfunded tranches of mortgage backed securities and CDOs. The data on mortgage backed securities and CDOs that is reported almost never includes unfunded tranches, because these tranches are more comparable to insurance contracts than to bonds. I include them in my estimate of triple A “assets”, because these insurance contracts can result in losses that appear on financial institution balance sheets, just as if they were funded tranches of CDOs.

Credit default swaps were used in two ways to increase the quantity of mortgage related securities available to investors. CDOs sold credit default swap protection on mortgage backed securities in order to increase their exposure to the mortgage market and CDOs bought credit default swap protection from highly rated financial institutions in lieu of selling their senior-most tranches to cash investors. (Some mortgage backed securities did the same on mortgage loans, but since the process was most commonly used by CDOs, I will refer only to CDOs from here on.)

To understand what was going on, you need to realize that mortgage securities were the hot product of the mid-naughties. Mortgages, and especially high-yield mortgages and mortgage backed securities, were in such demand that the average structured financier simply could not find enough for the CDOs he wanted to ramp up. In 2005 innovation in the credit default swap market opened up the possibility that a structured financier could create a CDO that was constructed of 10% mortgage backed securities backed by real loans (known as cash assets) and 90% credit default swaps (known as synthetic assets). Credit default swaps make a payment when the mortgage backed security referenced by the swap defaults. The investor was taking on a risk comparable to a 100% cash CDO, but the financier only needed to purchase one-tenth the quantity of mortgage backed securities he had had to find for a comparable CDO in 2004. The other nine tenths of the investors’ exposure came from using credit default swaps to reference mortgage backed securities that had already been packaged into other CDOs. Credit default swaps on mortgage backed securities made life a lot easier for structured financiers.

One consequence of the use of synthetic assets is, however, that more than one investor has exposure to each subprime mortgage backed security and thus to each underlying subprime loan. Another consequence is that structured financiers did not need cash to buy 100% of the assets in the CDO. Most credit default swaps are at least partially unfunded. This means that the seller of CDS protection promises to come up with cash in case of a default, but does not post cash to guarantee his capacity to do so when he signs the contract. When a CDO included synthetic assets, the structured financier only needed to raise enough cash – or sell enough tranches of the CDO – to cover the costs of the cash assets. You may ask, “But what’s the point of creating this huge CDO, if you’re only going to bother to fund 10% of the tranches?” The other 90% of the CDO, specifically the last-to-take-losses tranche of the CDO, could be “sold” to a large financial institution, like AIG or a monoline insurance company, or even held at the originating bank (e.g. UBS). That is, the financial institution would sell credit default protection to the CDO to cover any losses that were not covered by cash investors.

Whenever you hear someone talk about a “super senior” tranche of a security, this is precisely what they mean – some financial institution has written a credit default swap promising to pay up if defaults exceed the amount covered by the cash investors in the security. Super senior tranches of mortgage backed securities and CDOs are almost never funded.

If it happened that a structured financier had a CDO with more cash investors than cash assets, the extra cash was stored in “safe assets”, such as Treasury or agency securities or guaranteed investment contracts (GICs) issued by large insurance companies.

Now, perhaps, you begin to understand how it is possible for $2.6 trillion in mortgage loans to back $4.1 trillion in AAA securities. Quite a few of those securities were never funded and are really just promises to make a payment when there is a default on a referenced mortgage.

It’s also important to understand that the cash investors in CDOs – those who believed they were buying a “mortgage backed bond” – are going to be the first to lose their money. Why? Because there are many different classes of AAA securities – and it is the financial institutions that sold super senior credit default swap insurance that have a senior claim to any loans that do not default or that have significant recovery. The cash investors in every case bought subordinate AAA CDO securities that are guaranteed to be wiped out if defaults rise beyond a certain level. (Diagram 1 is drawn to scale, so that you can see visually the fraction of losses that will wipe out the subordinate AAA tranche.)

Notice something else about the consequences of this CDO structure given the housing crisis. As it is the super senior counterparties who have first claim to the underlying value of the mortgages, all that underlying value is likely to do is to reduce the amount that the super senior counterparties have to pay out in “insurance”. Since these counterparties initially assumed that there was almost no likelihood that they would make any payment at all, they were never prepared to pay out the whole notional value of the swap.

Thus, when spectators assume that the losses can’t really be too bad, because after all these are assets backed by real estate and there will be some recovery, they are failing to understand the nature of the ABS CDO market: (1) because the CDOs were built on high yield MBS tranches, their value can be wiped out entirely if sub-prime losses exceed 25%, Alt-A losses exceed 10% and Prime Jumbo losses exceed 3%; and (2) the super senior counterparties did not provision for making payment on their obligations. Whereas the cash investors have already put up cash that may be lost, the super senior counterparties will need to come up with the cash as defaults are recognized. Thus the super senior counterparties face not only losses on the balance sheet reducing their equity, but also a liquidity squeeze as they are forced to make payments on more than $1 trillion of swap obligations.

AIG was the first to be caught by this liquidity squeeze, but the monoline insurers are still struggling to manage the problem; the Swiss National Bank stepped in to save UBS from its embarrassments; and Citigroup had to take some super senior swaps on balance sheet in 2007 when its asset backed commercial paper conduits began to implode. Furthermore, a complete breakdown of the financial institutions exposed to super senior cash calls has yet to be made public.

Secondly, some mortgage backed securities have the same CDS based super senior structure as CDOs. While I have only located one explicit example of such an MBS (BSARM 2006-2 discussed in “Bear Stearns Quick Guide to Non-Agency Mortgage Backed Securities”) and have found no data on this issue, it was widely recognized that super senior MBS was being issued. Ominously, BSARM 2006-2 had a super senior tranche that amounted to more than 90% of the total value of the MBS. In diagram 1, I have guesstimated that $0.5 trillion of super senior swaps on MBS were issued. This would mean that in addition to the funded MBS on which we have data, there was another 20% of unfunded MBS issued.

Diagram 1 shows my estimate of the AAA assets backed by the $2.6 trillion in mortgages loans that got packaged into mortgage backed securities from 2005 through mid 2007. Diagram 2 shows the consequences of 25% losses on subprime loans, 15% losses on Alt A loans and 3% losses on prime jumbo loans.

The basic story is explained clearly in Fabozzi. My estimate of super senior ABS CDOs backed by MBS is derived as follows: From SIFMA we know that $685 billion in funded structured finance CDO tranches were issued from 2005 through 2007 Q2. I assume that 80% of the structured finance products backing these CDOs were MBS (see Fabozzi p. 175) giving us $540 billion of “mortgage backed” funded CDO tranches. I then assume that 70% of the ABS CDO structure was the unfunded super senior tranche (see Fabozzi p. 142) resulting in a total – i.e. funded plus unfunded – "mortgage backed" ABS CDO issuance of $1.8 trillion.

CDS "spread trades" and bubbles

A month or so ago there was some discussion of the "end of the world" trade. People have been trading credit default swaps (CDS) on the United States – these swaps are basically insurance against the possibility that the US defaults on its bonds – and to no one’s surprise the premium on this insurance has been going up lately. The trade in this product bewilders many for the simple reason that if the US were to default on its bonds, financial markets would be in such disarray that there is no reason to believe that any major investment bank or other seller of protection would be able to honor the insurance policy. That’s why CDS on the US are called the “end of the world” trade.

In response to the criticism of the trade, several financial commentators observed that these are in fact spread trades. (Here’s a particularly snarky example.) Because the premium is expected to go up, say from 0.7% to 0.8% over the period of a month, anyone who buys the contract today with the 0.7% premium will be due some collateral from the protection seller if the premium does in fact go up to 0.8%. (On the other hand, if the premium falls to 0.6%, the buyer of the contract will have to post collateral.)

Thus, someone who’s buying the CDS for a spread trade wants to receive collateral in case the premium on the contract goes up. One reason to do this would be to hedge: If you own Treasury bonds and are worried that interest rates will go up due to default risk and reduce the value of the bonds – the CDS can protect you from posting a mark-to-market loss on your balance sheet. But there’s a problem with this logic – interest rate swaps are a better way to protect against this possibility because they will compensate for a fall in the value of the Treasuries, whether or not the fall is due to a change in default risk. Furthermore the market in interest rate swaps is extremely deep, so the cost of entering into the interest rate swap is likely to be less than that of the CDS.

So what’s the other reason someone might buy a CDS on the US for a spread trade? To speculate on the movement of the CDS premium – if the premium moves as predicted, you can take advantage of the new premium by selling a contract at a higher premium than you are paying for a comparable insurance policy. Theoretically if there is a default requiring payment, you can simply pass on the payment that you receive from the bank that sold you protection. (Of course, this is the “end of the world” trade, so you’re probably thinking that if a default takes place, you may not get paid by the bank and you may not be able to honor your own obligations, but presumably this is such an apocalyptic scenario that it’s not really worth worrying about.)

I think the important question to ask is: What do spread trades on CDS on the US tell us about the market? They prove that there is a bubble in this CDS market.

People are trading something that they know has no fundamental value, in hopes of reaping short-term gains. Of course, the fact that the contract has no fundamental value means that someone will be caught at the top of the bubble, paying a high price for a valueless contract.

One phenomenon that supports this bubble is that, if the premium does rise, the value of the contract will go up and the buyer of the contract will benefit from short term accounting profits. The fact that everyone knows that this profit will disappear over the long-term, is a sign that these firms are “gambling for resurrection”. If they can make it through the short-term without declaring bankruptcy, they hope that longer-term investments or gambles will pay off allowing them to survive the crisis. On the other hand, they’re paying a premium today to preserve that possibility.

Who is selling the CDS on the US? Anyone with the balance sheet capacity to carry the interim losses that will have to be posted to the balance sheet if the premium goes up.

So, yes, it is true that “CDS on US government debt are spread products,” but this fact is itself a sign of how very, very sick our financial market are right now. It’s also proof positive that CDS are not always a good way to price default risk. The “end of the world” trade is a bubble and when it pops someone is sure to end up posting losses.

An invitation from the Banking Lobby: Enter the Matrix

I've been reading Congressional testimony on derivatives and I couldn't help but imagine the initial sales talk that took place some time before the action in The Matrix.

While credit derivatives are often pejoratively described in the media as a “bet”, it is important to realize that one could equally describe all investments as “bets”. When we buy the stock of a corporation, we are “betting” that the stock will be worth more in the future than what we paid.

The matrix has been carefully designed to simulate the world with which you are familiar. Any differences you perceive are not actually differences ...

The fact that the underlying credit is NOT a party to the agreement and, further, that neither the protection buyer nor the protection seller needs to own the debt of the entity, has also recently been subject to a great deal of media hyperbole. This fact is frequently and shrilly cited as evidence that credit derivatives are a “bet”. But exactly the same statement could be made about futures contracts or stock options – neither the purchaser or the seller of those contracts needs to hold a position in the underlying commodity or stock

You may be familiar with the matrix already in the form of an alternate reality games: there is really no need to distinguish between the matrix and harmless entertainment.

Consider Bank B which wants to diversify its credit exposure but does not have a relationship with the quality of credits it desires. Bank B can sell protection through a credit default swap as an alternative to making loans or buying bonds. This is economically equivalent to lending directly to the desired credits. [Read: While we don't actually lend any money, our derivatives are "economically equivalent to lending".]

When you enter the matrix we can assure you that you and your constituents will not notice that anything has changed.

It is important to remember that credit default swaps, like all derivative contracts, are zero sum contracts – the loss of one party in the contract exactly equals the gain of the other party. In aggregate, therefore, the losses incurred by protection providers equal the gains realized by protection buyers, making the overall CDS market a “closed system”, where gross losses equal gross gains, and both, when added, net to zero. This is in contrast to the cash bond market where credit losses result in permanent loss of value. [Read: Since it is unimaginable for a counterparty to go bankrupt, we have successfully created a financial product with no possibility of "permanent loss".]

Moving to the matrix brings with it great advantages. In a world without gravity, you can jump tall buildings at a single bound ...

The actual number that we should focus on is the gross replacement value of all outstanding credit default swaps, which according to the BBA was a little over $2 trillion at the end of 2007, or just under 3.5% of the notional amount for that period. That number represents the cost of replacing all the existing contracts in the market, just as the market price of an equity security represents the price at which it can be bought or sold in the open market. It is equal to the difference between the present value of fixed-rate premium payments to be made by protection buyers and the present value of the credit event-driven payments that the market expects will be made by protection sellers over the life of the swaps. [Read: Since everybody knows that we have perfected the models used to determine expected losses, you can feel confident that our prices are at least as accurate as simple stock market valuations.]

... or even dodge bullets.

In fact, according to the BBA, dealer positions represent more than 50% of the
of the credit default swap market and, as can best be determined from public
disclosures, have nearly equally balanced CDS exposures, consistent with the
dealer business model.

What? You're worried that when you and your constituents are all in the matrix, you'll be too dependent on the machines. Don't be ridiculous. As best can be determined from public disclosures, the machines are behaving consistent with our expectations. What could possibly go wrong?