Yet More Pirates -- CDS's

In today's Financial Times, Willem Buiter has an entertaining and disturbing description of how credit default swaps can be manipulated to effectively defraud investors and how one U.S. company has succeeded in doing this.

The obvious problem with CDS's, as Buiter recognizes, is that the purchaser need not have an insurable interest in what is being insured. The result is that

the CDS market is first and foremost a betting shop. You can buy CDS written on a given class of bonds, in amounts well in excess of the total face value of the bonds you own in that class. Indeed you may not own any bonds in that class and still buy CDS that pay off in the event a default occurs on bonds in that class. That is, CDS can be used not to hedge risk you already are exposed to, but to take on additional risk.

Unlike a pure game of chance, however, with a CDS, as with any type of insurance, information asymmetries are present, and as a result moral hazard, is a constant possibility. Moral hazard occurs when one party to the insurance contract -- traditionally the purchaser -- has information that the other partner -- traditionally the seller -- does not or is able to influence the likelihood of the insured event occurring. In other words, one individual may purchase life insurance on another knowing the other is gravely ill, or knowing that she or he plans to bring about the insured's demise.

Thus CDS's are unique instances in which moral hazard and lack of insurable interest can coexist. And as Buitner points out, this means that

[i]n the CDS world, the most common reported abuse of the instrument involved a party that owned both CDS and the underlying security, but had a net short position in the security. To be precise, assume I own $X worth of bonds of type j (at face value) and have purchased CDS on bond j that will pay out $Z if default occurs. If X < j: I am better off if default occurs.

Now assume that, as a bond holder, I can influence the likelihood of a default occurring. A possible scenario is where the company that issued the bond is in dire straits, but has a good enough chance of recovery and survival, that, from a social or economic efficiency perspective, it is undesirable to incur the real resource costs associated with a default. Assume the issuer of the bond has asked the holders of the bond to roll over the bonds, or to voluntarily extend their maturity. All bondholders but me have agreed. I am the holdout and the veto player. By refusing to go along with the voluntary restructuring (which, by assumption, would not be an act of default), I now can trigger a default, making a gain of $(Z-X). It’s socially inefficient; it may cause unnecessary human misery, but it is profitable and so, as homo economicus, I do it. Because of my hold-out position, I can drive the probability of default to unity, or 100 percent. This is the mother of moral hazard.

But now comes the mother-in-law of moral hazard. This time it is not the purchaser of the CDS (the insured party) who is afflicted by extreme moral hazard, but the writer of the CDS, the insurer. There is asymmetric information, but the informational advantage is with the insurer. Assume there is an amount $X of some bond of type j outstanding. Assume that the issuer of the bond is generally considered to be at significant risk of default. I now write (sell) CDS on that bond. Because there is no limit to the amount of CDS I can issue as long as there are willing takers, I can sell CDS to anyone who wants to have a flutter on the default of that bond. If I price my CDS aggressively (accept a low insurance premium per $ of bond j insured), I may be able to have a revenue from the sale of these CDS, $R, say, that exceeds the face value of the total stock of bond j outstanding. This would only happen if the total notional value of the CDS I sell (the total value they would pay out in the event of a default on bond j) is a multiple of the face value of bond j outstanding.

Having received revenue from the sale of CDS written on bond j well in excess of the face value of the entire stock of bond j outstanding, I then buy up, at a price above the prevailing market price (if necessary at face value or even above it!), the entire outstanding stock of bond j. As long as I can be sure I have the entire stock of bond j in my possession, I can be sure than no event of default will ever be declared for that bond. I, the writer of the CDS on bond j , and now also the owner of the entire outstanding stock of bond j , could simply forgive the debt I just acquired. The insurer has, ex-post, reduced the probability of default to zero. Those who bought the insurance (bought the CDS), wasted their money (their insurance premia).

Instead of buying up the entire outstanding stock of the bond directly and holding the bonds to maturity without calling a default, or forgiving the debt, I could instead, if the bond were some asset-backed security, purchase enough of the assets underlying the bond at prices in excess of their fair value to ensure that the issuer of the bond would have sufficient funds to pay off all the bond holders, should the bond be ‘called’, that is, retired prematurely. If in addition, I could make sure that the bond would indeed be called, I would again, through this financial manipulation, have reduced the probability of default on the bond to zero.

And indeed, a small U.S. brokerage, Amherst Holdings, has succeed in doing just this. It has

. . . sold CDS (default protection insurance) on mortgage bonds, then purchased the property loans underlying these bonds at above-market prices to prevent a default that would trigger payments to buyers of the contracts.

Some mortgage bonds can be “called,” or retired early, when the amount of loans backing the debt is reduced to certain levels by refinancing, loan repayments or defaults. The mortgage bonds targeted by Amherst fell into that category. So the mechanism through which Amherst made sure enough money would be available to the issuer of the mortgage bonds to pay the obligations due on these bonds, also caused the bonds to be called. The bonds were paid off in full, and the CDS Amherst had sold on these mortgage bonds became worthless.

As Buitner recognizes, the solution to this is obvious -- it is to require that the purchasers of CDS's have an insurable interest. After all, I cannot buy insurance on any future car accidents or house fires you might have. The potential for manipulation and abuse is just too great.