By R. Russell Hurst, Banc One Capital Markets
The answer may be shocking, but many investors have had healthy overall returns on seemingly large holdings of CDOs. On a strictly intuitive level, a 10% to 14% equity return on a stressed CDO, originally projected to earn 19% to 25%, compares very favorably to the performance of the stock market over the past two years. Some investors have been able to avoid the CDOs that severely deteriorated over the past five years, and have achieved positive returns at each of the Senior, Mezzanine and Subordinated levels. The question becomes, "How did they pick the winners and avoid the losers?" The answer can be found by looking more closely at the credit performance of the CDO market compared to the corporate market and the selection process used by most of these investors.
Despite all of the recent research by the rating agencies and others on the rating migration of CDOs, we found two supplementary statistics that help put the recent spate of CDO downgrades in perspective.
Over the past five years, 18.29% of all CDOs rated by Moody's Investors Service have had one or more tranches downgraded versus corporate downgrades of 19.68%. If the number of corporate ratings withdrawn is considered as a downgrade, the percentage increases to 41%.
More revealing, however, is the number of CDOs with multiple downgrades - 2.5% of CDOs (by issue) account for almost 10% (by tranche) of downgrades and 15% account for 37% of downgrades.
These statistics imply that it was possible for an investor to outperform the corporate market over the past five years by investing in CDOs. In other words, if an investment were made in a diversified static portfolio of corporates over the past five years, an investor in all the tranches of a particular CDO would, on average, have experienced fewer downgrades than corporates. Since a CDO at the equity or subordinated level represents a levered return on a corporate portfolio, collateral management or issue selection for CDOs did make a difference, albeit a small one. If we remove 15% of the worst performing CDOs (measured by multiple downgrades) the odds of outperforming the corporate market (from a downgrade perspective) greatly improve. Those CDO investors that have done well during this stressful period attribute their success to collateral manager selection, vintage and CDO product diversification, and the imperfections of rating agency models used to rate CDOs. These topics are discussed in more detail below and offer some insight to the process used by some investors to achieve superior results in the CDO market. Additional important aspects of our findings include the following:
* CDO market performance, from a credit perspective, was no worse than that of the corporate market
* 82% of all CDOs rated by Moody's have not been downgraded through the first half of 2002
* 87% of CDOs downgraded over the past 5 years can be attributed to those backed by HY Bonds
* CDOs, synthetic and cash, backed by ABS, Commercial Real Estate or post-1998 emerging-market collateral have had superior performance during this period.
* CLOs, synthetic and cash, with lower defaults and higher recoveries have done better, especially in the senior tranches, than HY CBOs.
* The use of sophisticated credit modeling tools has enabled many investors to outperform the market.
* Tail Risk may account for the inordinate amount of multiple downgrades experienced by some CDOs
CDOs versus corporates
The Rating Migration Studies of Moody's, Standard & Poor's and Fitch Ratings also show there has been room to successfully maneuver in the CDO market and achieve decent returns. Almost all of the multiple downgrades above can be attributed to those CDOs backed by high-yield bonds (87% per Moody's). Regardless of the downgrades, all ACF-CBO tranches except those rated Baa3' outperformed corporates with similar ratings. All but the Baa2' and Ba2' tranches of U.S. dollar synthetic balance-sheet CDOs outperformed corporates as well. CLOs, with about two-thirds of the issuer default level of the high-yield bond market and significantly higher recoveries, beat corporates across the board.
We were able to estimate that approximately 19.68% of corporate issuers with debt outstanding and rated by Moody's at the beginning of 1997 had been downgraded at least once over the five-year period ending in 2001. Further, a total of approximately 41% were either downgraded or had their ratings withdrawn.
The 19.68% downgrade statistic shows us what we might suspect, that the collateral managers of CDOs modestly outperformed the corporate market during this period. Without more information, we believe this statistic implies a better than apparent performance by the collateral managers of CDOs.
Manager selection
Our data shows that 82% of CDOs rated by Moody's have not been downgraded. Many one-time issuers with limited experience in the collateral class backing a CDO have produced disastrous results and have since exited the CDO market. Some issuers even though otherwise experienced at managing a particular asset class, settled on a strategy that went for yield regardless of implied credit risk. Other CDO managers pursued a sector strategy, such as telecom or early emerging market, and, even though the portfolio otherwise met the investment rules of the CDO, these collateral pools deteriorated beyond the point of repair. Successful CDO investors look for a winning track record from the manager in the collateral asset class backing the CDO and a compelling reason for the manager to be committed to that asset class over time.
Vintage and diversification
CDO investors that have diversified their CDO holdings by year of origination and in some cases, collateral holdings across all their CDOs by year of origination have also done well. The poor performance of CDOs originated in 1996, 1997, and 1998 is well documented. Diversifying your CDO investments or your aggregate collateral pool by year of origination will reduce the volatility of expected returns.
Another tactic for success in the CDO market has been to invest in the broad array of collateral asset types that are now offered by the structure. There have been few if any downgrades of CDOs backed by multi-class ABS and Commercial Real Estate. Corporate IG CDOs that contained one or more of our flock of fallen angels were downgraded but stabilized at the lower rating level and for the most part did not enter the downward spiral that has plagued HY CBOs. HY leveraged loans have experienced roughly one-half to two-thirds of the downgrades that hit the HY bond market and, due to their senior position in the capital structure and their secured nature, have had a significantly higher level of recoveries. As a result, HY CLOs have outperformed HY CBOs.
Understanding the models
The approach used by each of the rating agencies to rate CDOs creates a situation where once a deal gets in trouble it is likely that multiple downgrades will ensue. Both cash and synthetic CDOs in their simplest form receive tranched (usually 1st, 2nd, and 3rd) loss protection on a portfolio of assets with a certain expected loss. Ratings, representing a certain probability of default or expected loss, are assigned to each tranche. If the rating methodology is imperfect, certain tranches of CDOs with the same rating as others will have a disparate number of downgrades. If we were able to measure the expected loss and the distribution of that loss with precision, there would be no disparity. However, because this process is not yet precise, inefficiencies have created opportunities in the CDO market for investors.
Successful investors in CDOs have used credit models to view whether any CDO has been tranched or priced appropriately for the risk and whether each tranche has been rated in the right category by the rating agencies. The CDO, from that investor's relative value viewpoint, becomes rich or cheap, depending on whether default correlation is higher or lower than that used to model the deal.
It should also be noted that default correlation is not static, as it increases for the same portfolio during periods of economic stress. The rating agency models simulate stressful cash flow scenarios. But other than in this manner, the expected loss distribution does not flatten out in these stressful periods, as it should, to measure whether the CDO tranches have been properly sized around the now-changed expected loss distribution. This observation alone may account for the inordinate amount of multiple downgrades experienced by some CDOs.
