Credit Risk Management Failures: Is the Sky Falling?

After three modifications to their respective RMBS and CDO loss assumptions each agency has staked their positions as to worst case losses and probable losses. Moody's Investors Service (MCO), not surprisingly, assumes the highest losses in its rating methodology for bond insurers.

For MBIA (MBI), worst case loss on their entire book of risks (municipal and asset backed) is placed at $13.7 billion, probable loss at $4 billion, both in present value terms over the life of the risks (Rating Action 2/26/08). For Ambac (ABK), on the same basis, Moody's assumes $12.1 billion worst case loss and $4.1 billion probable loss (Research Announcement 2/29/08).

The Rating Agency stated that the existing capital of both insurers exceeded the minimum level for a Aaa claims paying rating. MBIA is short $1.7 billion and Ambac is short $2 billion against the Moody's target comfort level of 1.3x worse case loss for an Aaa claims paying rating. Holding company ratings have always been lower than insurance company claims paying ratings.

The worst case or stress case scenarios that the rating agencies use are in fact depression (1930's style) scenarios. Simply put, at any point in time a triple-A financial guaranty insurer must have minimum capital equal to 100% of depression losses at a 99.9% probability with an extra 25-30% capital cushion thrown in for good measure. No one who understands the business has ever suggested that an insurer would necessarily maintain a triple-A rating during or after a depression, but they would have sufficient capital to pay their obligations.

The 25-30% capital cushion provides a margin for error and remaining capital to support the insurer as a going concern post depression. All of the rating agencies assume that a depression can begin at any time and will last about six years. The agencies all use the same basic methodology but can differ on worse case loss assumptions for different kinds of risk.

The rating methodology applied to bond insurers has evolved but has not fundamentally changed over the past 30 years. Loss assumptions, however, for RMBS and CDO have been dramatically increased, as reflected in declining capital adequacy for bond insurers and rating downgrades on uninsured and insured RMBS and CDO debt.

Using Moody's as the benchmark, Ambac and MBIA, for instance, have seen their capital adequacy ratio decline, after capital infusions of $2.6 billion for MBIA and $250 million for Ambac, to 1.2x and 1.13, respectively. Essentially, since June of 2007, MBIA has been assessed an additional capital charge of $4.7 billion ($2.6 billion raised with 1.7 billion to go). Ambac has been assessed $2.25 billion ($250 million raised with $2 billion to go). All as a result of insuring RMBS and CDO's via CDS (credit default swaps).

Ambac had a higher capital adequacy ratio than MBIA as of mid year 2007.

What went wrong and are the rating agencies' assumptions correct? The failure of bond insurer and rating agency credit risk management was a gross underestimation of loss assumptions, to state the obvious. Both parties failed to recognize or chose not to fully recognize the dramatic decline in mortgage underwriting standards that began 2004.

Both embraced more complex versions of CDO structures such as CDO of CDO where pools of CDO's were packaged and loss collateral was sized, in part, on the erroneous assumption that each underlying CDO pool would perform about the same. In these cases, not only were loss assumptions inadequate, the structure itself was defective.

Just as important, the rating agencies and participating bond insurers failed to fully appreciate the correlation that is unique to single family loan securitizations. Correlation, the degree to which individual assets move in concert, exists in all securitizations in varying degrees and in type. Correlation based on location (geography) is not a driving factor in credit card securitizations but it is when it comes to residential real estate securitization. The rating agencies and bond insurers have "adjusted" their views on correlation in residential real estate securitizations.

Are the current rating agency assumptions and resulting ratings reliable after having been asleep at the switch? With franchises and reputations at stake and share prices down about 50%, I would say yes. At a minimum, they are either calling it as they see it or erring on the conservative side. The sky is not falling.

If you found this article interesting, you may want to read my last piece, Ambac MBIA, the Shorting Game is Over.