Friday, October 17, 2008

Proposal on Transforming CDS: Credit Insurance Trust

The latest round of eruption of year-long financial volcano has thrust CDS into almost bar-conversation vocabulary. The common sentiment seems to be total hatred or disgust. Eulogies for CDS have already been published. Others are calling for change in how CDS is traded and/or treated in accounting. Common proposals include

1. Regulate CDS as what it is, insurance.
2. Trade CDS on exchanges so as to minimize counterparty (seller) risk.
3. Move CDS on balance sheet.

But the cost to the seller -- capital reserve in 1 and 3, margin collateral in 2 -- would be prohibitively high if the capital/collateral cushion is to be enough to meaningfully reduce counterparty risk. Such high cost to the seller translates to high premium for the buyer. Result: dramatically reduced market size. In other words, CDS is still dead.

Few people would shed a basis point of a tear on such news today, I suspect. But let's not forget CDS can serve a legitimate and important function: hedge. Yes, there are real, meaningful hedges using CDS. If you hold a bond or otherwise are exposed to credit risk of somebody, there's nothing like CDS that can provide direct, fast, efficient, and clean hedge in time of need.

Credit derivatives are NOT financial weapons of mass destruction. They merely have the capacity of being such powerful tools. But whether it serves good or evil depends on people. Blaming financial products strikes me as profoundly misguided and ignorant, even dishonest.

No, credit derivatives didn't cause this crisis. WE -- the government, some market players, some mortgage lenders, and some home buyers -- caused this crisis. We didn't use the powerful tool of credit derivatives properly.

And while we're on it, let's make one point very clear. The reason why we failed to use it properly is not due to lack of understanding of the risks; rather, it's because the system -- regulators, laws, and corporate governance -- has incentivized decision makers at all levels to ignore the risks and focus on short-term gain. Calling regulators and Wall Street executives stupid or incompetent might feel good. But it wouldn't be correct at the overall level.

It's the system that's skewed and flawed. The magnitude and duration of decision makers' risk and reward are misaligned; sometimes even the sign is wrong. If we only focus on a few individuals' misjudgment or unethical/criminal behavior, we run the risk of missing the deeper, greater cause -- the systemic flaws. In doing so, we only set ourselves up for a repeat in the future.

So I'd like to focus on addressing the systemic flaws here, and only on CDS. Let's first take a look at what makes CDS unique -- forget about the standard pricing model or how it's been treated so far, focus on the economic idea behind it.

1. CDS by nature is an insurance.

2. However, there is one important distinction between CDS and traditional insurance products. The latter can be "hedged" via diversification. It's very difficult to hedge CDS this way because, by definition, it deals with low-probability, high-weight events among a small population. This is the most fundamental flaw of CDS as we know it today. It goes right into the well-known fact that statistical applications (not the theory) loses relevance as you migrate towards the distribution tails. No company is big enough to provide meaningful insurance of such risks on a meaningful scale, especially since the company is a credit risk itself. As to governments playing any role, I hope the governments' involvement in this crisis has thoroughly disgusted everyone so let's not even go there.

3. CDS differs from most of other derivatives in that it is inherently highly leveraged. This makes CDS the key ingredient of the chain reaction.

If you understand the above three fundamental characters of credit default protection, you'd realize regulation or accounting gimmicks could not possibly address the real cause of the problem, short of killing the market altogether. The real solution must address all of the three fundamental characters while avoiding throwing the baby out with bathwater.

In addition, just as nobody should, in the legal sense, moral sense, or from societal considerations, reap a windfall from tragedies, as is the case for all traditional insurance, nobody should reap a windfall from default through CDS. The social utility of insurance is to lessen the impact of tragedies as opposed to encourage speculative or reckless behavior. You may be tempted to ban naked CDS based on this consideration. But naked CDS has its legitimate use, since all exposed to a credit risk do not hold the bond.

More importantly, since CDS settlement is by definition a zero-sum game, there is absolutely no reason the society as a whole should be greatly damaged by it aside from the fact that the reference entity has just defaulted. If we cannot find a way out of this PURELY artificial problem without sacrificing its beneficial functions (or going into state capitalism as we are), then I don't know how we deserve the top spot on the food chain.

My proposal is to set up a private, non-profit Credit Insurance Trust (CIT). To begin with, let's call it what it is, Credit Insurance (CI).

1. Those who want to sell CI must contribute capital, the amount of which is tied to the total amount of notional they can sell. The relationship between contribution and allowable notional is determined by auction, e.g., $1 contribution gives you a permanent, revocable, transferable license to sell up to $1000 notional CI (hold you protest on such outrageous leverage -- read on). Licensees are subject to trustee approval. Licensees are seller agents for CIT, which is the legal seller of all CI. Licensees can buy additional licensed amount at prevailing auction price, subject to the total cap set by the Board of Trustees.

2. Board of Trustees is elected annually by CI buyers. The vote is weighted by notional outstanding bought.

3. CIT invests the fund in long-dated treasuries (or other "riskless" securities the BOT deems appropriate).

4. CI premium goes to the Trust. Licensees take a haircut, ranging from 0 to 5 bps. The better the credit, the bigger the haircut. This discourages speculative selling on poor credits. The returned premium goes into a separate fund -- let's call it CITIF -- which also invests in "riskless" securities, but with maturity no longer than one year. Furthermore, CI on a credit cannot be sold once the credit quality deteriorates beyond certain level. This prevents agents from selling garbage while making fees via private arrangements. Existing CI contracts can always be transferred, however, with registration to CIT.

5. Returns on CIT and CITIF investment pay for the administrative cost first. Excess goes into CITIF; deficit comes from CITIF. It's to the BOT's interest not to let CITIF run dry.

6. CI buyers must invest all of the settlement windfall into CIT, which is done automatically as part of the settlement process. In return, they become holders of CIT's interest-only, non-compounding, annual coupon, 10Y bond. In other words, each settlement is automatically a CIT bond issuance. Coupon payments of such bonds come from the accumulated funds in CITIF, set in arrears and capped at 20%, after administrative expenses. CIT bonds can be auctioned at issuance (buyer cash out) or traded on secondary market.

This may seem a bit complicated, and will get more so when put in the international context. Price of CI depends on the valuation of CIT bonds, the pricing of which would be quite interesting. But it has numerous advantages:

1. While credit default risk may have proven too high-weight for any particular company to bear, by definition such risks can be handled, and statistically "hedged" in the same sense of traditional insurance, at the macroeconomic level. This setup solves the fundamental flaw pointed above.

2. There's no economic incentive for speculative selling by the agents. Speculative selling of CDS is the biggest systemic flaw currently. This setup cuts off the chain reaction mechanism from the root.

3. Contrary to the destabilizing effect of CDS, this setup serves to stabilize the system and benefits everyone involved, therefore by extension the society in general. It will also lower the systemic correlation, which is extremely beneficial to many other credit derivatives, e.g., certain types of CDO.

4. Speculative buying is also greatly discouraged. As a credit worsens, the haircut by seller-agents decreases, discouraging them to sell protection on the credit. Also, buyers' payout is directly tied to the future health of CIT; it's to their interest to keep it well.

5. It aligns risk and reward perfectly. In essence, the insurance is backed by the pool of premiums, not by the promise of any single entity. Licensees make the fee for their service and providing the initial funding to the Trust.

6. CIT's fund is guaranteed to be perpetual. The only way it goes bankrupt is for its "riskless" investment to go sour. In such worst-case scenario, the government would have defaulted. The biggest risks cannot be hedged, period (one example is Earth being demolished by Vogons).

7. Buyers can remain anonymous until settlement. This is important to protecting their interest. But anonymity in this framework will not have the side effect of its current form because the net position on each credit can be easily calculated on the seller side, which is completely public.

8. Last but not the least, this setup is in fact extremely simple. It heavily relies on the market, minimizing reliance on government intervention or artificial, arbitrarily set rules and prices.

Let's run some numbers. Say CIT total fund is $1B. Based on auction price of license amounts, $1T notional can be sold. Average premium is 100 bps (1%), or $10B, per year going into CITIF. Average haircut by licensees is 2 bps, or $200M per year. In an EXTREMELY bad year, 10% of covered credits (by notional sold) default. That's $100B. Buyers still get 10% coupon for the first year. Since in such dreadful times people would want to buy more protection, and premium would have increased, future coupon on CIT bonds will likely increase. (I've omitted investment income and expense for simplicity.)

Everybody wins.

No comments: