Suppose I lend you £79 and you lend me 100 and we set each loan to be variable-rate at the respective central bank's base rate (4.5% ad 3.75%) respectively. We agree to pay each other interest each day.
Because the exchange rate today is 0.79 /£ no money need change hands when setting this up. We can also agree to put some collateral (say, you post 10 and I post £8) in escrow and if exchange rates fluctuate we will top it up by the difference between the nominal values of the two loans. If we can't keep at least the initial amounts in the accounts the contract gets cancelled at nearly zero cost because the notional values nearly match and there's enough money in the escrow to make up the difference.
The notional value of this arrangement is 200 at inception, but in fact only 20 are posted as collateral. The only risk is that exchange rates will fluctuate more that 10% in a day and the party on the wrong side of that movement won't have enough cast lying around. But the other party is covered for the first 10% from the escrow account.
Leaving aside the question of why should we want to enter into such an agreement (called a currency swap) how would you justify makig it illegal, and which amount is more representative of the quantities involved? 200 (notional) or 20 (collateral) or the average daily payment over time? A vivid image of what should exist acts as a surrogate for reality. Pursuit of the image then prevents pursuit of the reality -- John K. Galbraith
In the credit default swap, the amount to be paid in the event of default is not in escrow anywhere. And, of course, since the party acquiring the notional coverage of the CDS need not be holding the bond that is it covering against default, it is quite possible for the exposure to default via CDS to be many multiples of the actual assets of the firm.
While originally focused on the moral hazard, the requirement that an insurance policy can only be issued to someone with an insurable interest means a much lower ceiling on exposure to systemic risk in the insurance market proper than in the quasi-insurance CDS market.
Now, an uncertain part of this multiple exposure is laying off the exposure by balancing a CDS exposure with CDS coverage ... but as the Aon / Societe Generale (apologies spelling) case showed, small differences in the description event can lead to that presumed cover vanishing in the event.
Indeed, I can see an argument for treating CDS where the acquiring firm owns the bonds (or other asset) as insurance and regulating it as such, and treating naked swaps as speculations and leaving them in the YOYO basket. Utsukushikereba sore de ii
Here are the correct wiki links:
cavalerie
Derivatives