Advance rates are the crucial variable in market value deals. It is useful to look more closely at how these are derived. Advance rates are actu- ally a combination of three factors: price volatility of the securities, cor- relation among securities, and liquidity. It is interesting to look at how Moody’s and Fitch, the two rating agencies that have rated the bulk of the market value transactions, view each of these variables.
Both Moody’s and Fitch use historical volatility as the basis for deriv- ing volatility estimates. This volatility is then stressed depending on the length of the historical record and the desired rating of the CDO tranche.
Because there is a very complete record for the returns on high-yield bonds (that is, high-quality information collected over a large number of years) only a relatively small stress factor is applied to the historical volatility for this instrument. At the other end of the spectrum, a relatively large stress factor is applied to distressed instruments, especially reorganized equities.
The higher the desired rating, the greater the stress factor, which reflects the fact that higher rated tranches are expected to hold up under greater standard deviations of stress. Fitch is very explicit on that final point. A security rated A must be able to sustain market value declines three times as large as needed for the security to obtain a single-B rating. For an AAA rating, the security must be able to sustain market value declines five times as large as would be needed to obtain a B rating.
The choice of market value correlations is problematic. Historical correlations are useful, but correlations often rise sharply during peri- ods of crisis. Thus, Moody’s uses correlations that are higher than those prevailing during “normal” periods, but not as high as those observed during the most stressful periods. Moody’s assumes correlation of 0.55 between firms within the same industry, and 0.40 among those in differ- ent industries.
For most securities, the bid-ask spread is small relative to ordinary price volatility. However, market value transactions lend themselves to
c10-Market Value CDOs Page 212 Monday, March 6, 2006 11:19 AM
using less liquid assets that also have irregular cash flows. For these securities, liquidity can become a key consideration, especially during periods of financial stress. Both Moody’s and Fitch make assumptions as to what losses would be during periods of market stress. So, for per- forming high-yield bonds, Moody’s assumes a 5% liquidity “haircut,”
while for distressed bonds, its “crewcut” is 10%. For performing bank loans the haircut is 7%, while for distressed bank loans it is increased to 12.5%. Reorganized equities get scalped at 20%.
So these three factors—price volatility, correlation among securities and liquidity together—account for the advance rates shown in Exhibits 10.5 and 10.6.
Are Advance Rates Conservative?
To test how conservative advance rates actually are, we look at monthly performance for a readily available set of data—high-yield bonds. We used the performance of Citgroup’s High-Yield Bond Index (available on Yield Book). This data cover January 1989 to October 2005.
We use monthly observations because the market values are evaluated at least weekly, and a portfolio normally has 10 to 15 business days to liquidate assets to correct the deficiency. Thus, if the portfolio passes the test one period, and then the market value of the portfolio deteriorates, it could take a maximum of just over four weeks to find and correct the deficiency (1 week maximum until the next test, 15 business days to cor- rect the deficiency). This suggests that monthly intervals are the correct benchmark period for looking at how conservative advance rates are.
The index includes all publicly traded domestic debt with a fixed- rate coupon, a minimum maturity of one year, and a maximum credit quality of Ba1. The index excludes payment-in-kind (PIK) bonds, and Eurobonds.
There are 202 observations in the period investigated. Exhibit 10.9 shows the 1-month returns as a histogram. The monthly return distribu- tions are similar in that they are both representative of the market, and their returns are usually close. Note from this histogram that there was only 1 month in which the total return on the Citigroup index was less than −8% (the worst single month was July 2002 at −8.81%).
The worst three months on the indices were May, June, and July 2002.
The Citigroup index was down 1.24% in May, 8.81% in June, and 4.52%
in July 2002.
We showed earlier that the advance rates are meant to correspond to one-month price changes. However, the advance rates are more severe than the worst three-month period in the history of the market. Clearly, the advance rates are very conservative.
214 OTHER TYPES OF CASH CDOs
EXHIBIT 10.9 Monthly Return Distribution—Citigroup High Yield Indexa
a Monthly returns since 1/1989–10/2005.
Further Evidence of Conservative Aspect of Advance Rates
Want proof of the extent of conservatism in the rating methodology?
Market value CDOs have been one of the best performing types of CDOs. Downgrades have been minimal. Many investors will find that particularly surprising in light of asset price volatility over the past decade. Clearly, part of the answer is that the vast majority of CDOs have been of the cash flow variety. However, another part of the answer is that the advance rates are so conservative that price volatility in recent years is well within the range anticipated by the advance rates.
One-Month Return Range Yield Book
–9 to –8 1
–8 to –7 1
–7 to –6 1
–6 to –5 1
–5 to –4 2
–4 to –3 4
–3 to –2 5
–2 to –1 18
–1 to 0 20
0 to 1 49
1 to 2 59
2 to 3 27
3 to 4 8
4 to 5 0
5 to 6 2
6 to 7 2
7 to 8 1
8 to 9 1
9 to 10 0
10 to 11 0
11 to 12 0
Total Observations 202
Mean 0.73
c10-Market Value CDOs Page 214 Monday, March 6, 2006 11:19 AM
CONCLUSION
Many CDO investors have steered away from the debt in market value deals, believing that purchasing the debt is like making an investment in a hedge fund. As a result, market value deals trade at similar or slightly wider spreads than cash flow deals launched at the same time. The pro- tections built into market value deals are quite powerful from the bond- holder’s point of view.
c10-Market Value CDOs Page 216 Monday, March 6, 2006 11:19 AM
Five PART
Synthetic CDOs
p05 Page 218 Monday, March 6, 2006 11:19 AM
CHAPTER 11
219
Introduction to Credit Default Swaps and Synthetic CDOs
n this chapter we describe the basic workings of credit default swaps (CDS) and synthetic CDOs. These products have grown tremendously since 1996 in terms of both trading volume and product evolution. Spe- cifically, the volume of outstanding credit CDS rose from $20 billion in 1996 to $2.3 trillion in 2005. In 1997, $1 billion of synthetic CDOs had been created (including funded, unfunded, and CDS index tranches); by 2005 there were over $1 trillion of synthetic CDOs outstanding.
In terms of product evolution, CDS have developed from highly idiosyncratic contracts taking a great deal of time to negotiate into a standardized product traded in a liquid market offering competitive quotations on single-name instruments and even indices of credits worldwide. Synthetic CDOs have evolved from vehicles used by com- mercial banks to offload commercial loan risk to customized tranches where investors can select the names they are exposed to, the level of subordination that protects them from losses, and the premium they are paid (although not all three simultaneously).1 Finally, the rise of stan- dardized tranches on CDS indices has increased trading liquidity, thereby allowing long-short strategies based on tranche seniority or pro- tection tenor.