The first thing that's needed is homogenization. Currently CDS are liquid so long as there remains 5-years to termination. But as soon as a contract rolls to "off the run" liquidity disappears. That's a major reason why CDS contracts have ballooned to over $60 trillion in notional outstanding. No one actually terminates the contracts, they just buy offsetting contracts.
In addition, CDS are currently struck with various initial spreads. So one person might own Morgan Stanley CDS with a deal spread of 100bps, and someone else with 200bps, and another at 300bps. This also lends itself to illiquidity and poor price transparency.
Then there is the problem of defaults. When an issue defaults, the buyer of protection may deliver a bond to the seller of protection in exchange for par. Except when there are large-scale defaults, like in the case of Lehman or the GSEs, actual delivery of bonds is impractical. So they hold an auction to determine a theoretical value for all outstanding bonds, and that amount of cash is exchanged between sellers and buyers of CDS protection. The problem is that whole process creates a huge degree of uncertainty, and only lends to the feeling that CDS are too easily gamed.
But CDS wouldn't be hard to boil down to a very basic tradable contract. First you set all contracts with a 5% coupon paid quarterly. Second, the contracts have 5-year maturities, with new contracts created each year. Third, in the event of default, the seller of the contract pays the buyer 60 cents on the dollar. No actual bonds change hands. That's it.
The contract would trade based on the present value of the 5% coupons vs. the expected default probability of the referenced company. This may result in either the buyer or seller of protection making an initial cash payment to the other party. For those familiar with the vulgarities of CDS, it would be similar to how up-front contracts work now, except that it could cut both ways.
For example, take a relatively low risk company, say Johnson & Johnson. Let's say that you estimate the proper default spread given J&J's default risk is 0.6%. Since the contract stipulates a 5% annualized payment, the recipient of the 5% coupon (seller of protection)must make an initial payment to the buyer of protection. The opposite would be true for higher-risk companies, like General Motors or MBIA.
Perhaps the best part of this is that a simplier product could be more widely adopted. Sellers of protection would have a defined set of gain/loss scenarios if held to maturity. Currently CDS trade only among large institutions, but wider distribution would certainly improve liquidity and price transparency.
And contracts of this sort could be implemented by exchanges tomorrow. And we wouldn't need some massive new regulatory scheme for CDS. Just simplify the contracts and put it all on an exchange, and all kinds of problems just float away.
Now yes, we want to start winding down the massive number of CDS contracts outstanding, if for no other reason than to eliminate the systemic risk. That's easy enough. Tell banks that any non-exchange traded CDS contract has an additional risk weighting to account for counter-party risk. Banks will immediately start pairing off their CDS exposure and looking to replace it in the exchange-traded market. That would go a very long way to reducing current CDS notional outstanding in a very short period of time.
It can be done. Let's see if anyone actually wants to solve this problem or not.