S&P has finalized a revision of the rating methodology for bond insurers. Shares of MBIA (MBI), Assured Guaranty (AGO) and Radian (RDN) rallied on the news. An examination of the announcement reveals limited reason for celebration, since the most onerous provision, an excessive capital requirement for the AAA rating, remains intact.
When the original request for comment was released, bond insurers tanked on the news, since the proposed changes set insuperable barriers to the achievement of high ratings, and reflected a deliberate brand of stupidity designed to set the much needed dialogue on ratings back thirty or forty years. For a detailed discussion, the public comments (pdf) of AGO, available on its website, are extremely informative.
Briefly, S&P ignored its own findings going back to when it originally developed the ratings methodology and carefully misinterpreted the Hempel study on which any intelligent discussion must rest. To cap it off, S&P defined capital requirements in a way that ignored the UPR (unearned premium reserve), which is a large part of the claims paying capacity of any bond insurer.
The final version gives ground on the misinterpretation of the Hempel data, but holds fast on the excess capital requirement for the AAA rating, staying with a 75:1 capital ratio and refusing to consider the UPR. The reason cited is curious - that regulators have on occasion taken action even when there was substantial UPR. Rating is not about avoiding regulatory intervention, it's about being able to pay claims. It's difficult to see how or why a regulator would intervene in a way that would prevent to use of the UPR to pay claims. That's what it's for.
Here's the essence from S&P:
As discussed in paragraphs 38-40, in determining the final rating, the leverage test will act as a filter only for companies with 'aaa' adjusted indicative ratings. To achieve and maintain a 'AAA' rating, there is a maximum leverage level of 75:1 for all risk categories. The test is limited to 'aaa' indicative ratings and 'AAA' rated insurers because there is the possibility of stress associated with either model error or event risk that, although remote, is not otherwise captured by the 'AAA' stress scenario used in the criteria. Moreover, higher leverage would not be consistent with 'AAA' credit stability. Capital, the denominator of the leverage ratio, is defined as statutory capital and excludes loss and loss-adjustment expense (LAE) reserves and unearned premium reserves. Loss and LAE reserves are excluded because their inclusion would overstate the capacity of statutory capital to absorb future credit shocks. Consistent with "Request For Comment: Bond Insurance Criteria," published on Jan. 24, 2011, unearned premiums are not included in capital because there have been multiple instances where a regulator intervened despite the existence of significant unearned premium reserves. This test is focused on the possibility of regulatory intervention as opposed to an insurer's access to cash to pay claims.
Failure to Establish Constructive Dialogue
It isn't possible to have an intelligent discussion of the issues involved in rating bond insurers without developing a clear and factual understanding and agreement about what caused the meltdown of a majority of AAA rated bond insurers. There are certain issues that rating agencies, and investment banks, and regulators for that matter, just don't want to talk about.
I'm talking about dishonesty, misrepresentation, omission of material facts, and outright fraud. Insurance rests on the assumption that losses are random occurrences, beyond the control of any party to the contract. However, loss on a CDO of synthetic ABS stuffed with adverse selected CDS is not a random occurrence - it's a predictable, and carefully planned, outcome. And one to which the rating agencies were accessories.
Similarly, loss on RMBS where 80% of the collateral is ineligible because it does not meet representations and warranties made in order to complete the sale is entirely predictable; indeed, inevitable. And the rating agencies were complicit: they were in it for the fees and carefully played three monkeys throughout.
The original bond insurance business model was workable, and had functioned very effectively for thirty years or more. The failure of a majority of the bond insurers can be attributed to fraud, plain and simple. MBIA has stated that if it is successful in exercising its rights under representations and warranties, and in recovering losses caused by fraud on the part of sponsors and originators, it will recover all of the claim payments on MBS and CDO's of ABS.
The equations and calculations that underlay structured finance as it was practiced during the housing bubble were accurate enough. The problem was, that the collateral did not behave in the same way as it had in the past, because it was tainted and indeed polluted by fraud and dishonesty at every link in the chain of securitization. Until the fraudulent contracts are pushed back up the chain to the responsible party, and the various players have admitted (and made restitution for) their offenses, the whole system will remain treacherous and unstable.
Insurance underwriters are taught about the requirements of an insurable risk (or peril) at an early stage in their training. The concepts are basic, but important, and can shed some light on the issues surrounding bond insurance. Here are the requirements:
Chance of loss must be calculable by the insurer
Premiums must be affordable
Loss must be noncatastrophic
Large number of homogeneous exposures
Loss accidental from the insured's point of view
Loss must be measurable as to amount and number
Quite simply, the losses caused by pervasive fraud aren't insurable. The bond insurers never accepted these risks, and protected themselves by requiring representations and warranties. The banks, regretfully, have been very reluctant to honor their contractual obligations.
And the banks have artfully contrived to absolve themselves from charges of fraud because the contracts were written in such a way as to protect the insured (the persons or institutions who bought the bonds) from loss caused by fraud. The lines somehow get blurred, and the perpetrators claim protections that were intended for the victims.
Once fraud is removed, the remaining obstacle relates to catastrophe exposure, meaning the possibility that losses will not be spread out over time. Car accidents occur more or less randomly throughout the year, and are not subject to catastrophe. Homeowners claims can occur simultaneously from a cause like hurricanes or tornadoes, and the risk can only be made insurable by spreading it out by means of reinsurance.
The concern is, that bond insurance losses will spike during a recession or depression. A level of capital that will easily handle losses spread out over time may prove inadequate if the losses occur all at once. Because bond insurance losses are normally paid out over time, with the insurer making payments of principle and interest when due under the terms of the bond, this objection is somewhat ameliorated.
During the Depression, many municipalities missed payments on their bonds, but most of them eventually made good on their obligations. So the issue of adequate capital is largely about a temporary spike in claims, which will ultimately be reduced by recoveries. It's about confidence and access to capital. S&P is suggesting that the only answer to this difficulty is to hold capital well in excess of what will actually be needed in the worst case scenario.
It happens over and over. When the financial institutions that support the commerce of a nation come under stress and confidence disappears, the government gets involved, as guarantor and ultimately as savior. As a society, we will all be better off if we accept the facts of life and work to make the process as efficient and fair as possible. The FDIC has functioned well, and a similar concept could be applied to insurance, at least to catastrophes that exceed the capacity of the industry to resolve by means of reinsurance.
The best way of handling it is to keep the government out of it, except as an ultimate reinsurer, and to have the framework firmly in place before the trouble starts.
The question comes up, are municipal bond defaults caused by overall economic conditions, or do they arise from local considerations? The latest occurrence in Jefferson County, Alabama, was fueled by a heady combination of corrupt politicians, sleazeball contractors, and greedy Wall Street types. The recession had little to do with it.
S&P has said that it doesn't expect an epidemic of municipal bond defaults from the recession. If so, the catastrophe issue demonstrably is not as important is it appears.
If the catastrophe exposure, where all the towns in the US go broke simultaneously, is not critical, then the factual chance of losing money on an insured bond is a compound probability: both the municipality and the bond insurer must go broke. If there's a 5% chance the municipality will go broke, and a 5% chance the insurer will go broke, the chance of both occurring is 0.25%.
The Value of an Opinion
The rating agencies stress that their reports are opinions, and constitutionally protected as free speech.
The opinions of the rating agencies proved worthless during the financial crisis. They were worse than worthless: they were pernicious, enabling massive fraud and misallocation of investment assets. Investors who were looking for low risk investments were hustled into totally unsuitable junk. The US is still paying the price for this stupidity.
The elaborate edifice constructed on the weak and shifting sand of rating agency opinions is unstable, fragile, and a danger to the financial welfare of the nation. The whole system needs to be carefully examined and rebuilt from the ground up.
A contract of insurance from a well-capitalized insurer is better than a rating agency opinion. After all, the insurer will make you whole, to the limit of its resources. The rating agency will defend its freedom of speech, and pay you nothing.
S&P's announcement does not set the stage for a return to the good old days. The model of the AAA rated insurer sprinkling its customers with pixie dust may not work again. If the US government can't keep its place in the circle of light, why should bond insurers be admitted?
A bond insurer holding a large amount of excess capital to support a AAA rating would invest the funds in very safe bonds, presumably treasuries, with very low returns. Under the circumstances, it would be difficult to achieve a reasonable ROE. If a company earned 15% ROE on the part of the capital actually and necessarily used in writing insurance, and 2.5% holding excess capital in treasuries, the shareholder would have the unpleasant combination of a risk-free rate of return with a nasty dose of tail risk.
Under the compound probability reasoning advanced above, the sacrosanct AAA rating is not necessary in order to achieve a genuine improvement in the risk profile of an insured bond. And an insurance policy is better than an opinion.
I continue to believe that AGO and MBI have substantial room for share price appreciation, based on nonGAAP adjusted book values, and the expectation that both will eventually be able to write meaningful quantities of profitable new business. Time and patience are required. The ultimate structure of the credit enhancement business has yet to be determined, and the respective positions of insurers and rating agencies may require modification.
Moody's (MCO) and Standard and Poor's parent, McGraw-Hill (MHP), are unattractive investments, because they sell worthless opinions.
There are serious issues that require discussion and resolution. S&P as presently constituted appears to be incapable of taking a meaningful role in this process.