Guest author: Bob Selvaggio, with an introduction by Donald R. van Deventer
My role today is very brief: to introduce today’s guest author Dr. Robert Selvaggio. Bob asked me to describe him this way: “Bob is a long-time NYC bank and insurance company risk manager," but that’s far too self-effacing. Bob has been one of the most respected risk managers in the United States for more than 20 years. What follows are Bob’s personal opinions and not those of his employer.
Donald R. van Deventer
Observations on the Monoline Meltdown
The bursting of the housing bubble has left the monoline financial insurers in shambles, and it is unlikely that the financial guarantee industry will ever regain viability. Risk managers, CFOs and municipal treasurers have learned the hard way that the “rating agency capital” or “claims paying resources” the guarantors held in support of their triple-A ratings were very poor substitutes for formally documented collateral agreements pledging specific assets against potential counterparty exposures; they will not again repeat the same mistakes. The latest data based on actual trades put the value of policies previously written by the now junk-rated monoline insurers at about ten cents on the dollar owed (a November 18, 2009 article from The Street entitled “Ambac Misstates Financials to Meet Minimums” includes mention of a recent distressed exchange of a negotiated $520 million recovery against Ambac’s likely default on $5 billion of ABSCDO insurance claims). ABSCDOs are collateralized debt obligations formed from portfolios of asset-backed securities most of which are mortgage-related. As an informed former insider and observer of the financial guarantee industry over the past decade, I would like to share my own perspectives on its demise in order to help prevent future catastrophic losses to the shareholders, creditors, employees and clients of firms that seek to earn returns from active risk-taking.
Conventional wisdom holds that the monolines failed due to their aggressive and naive underwritings of insurance or credit default swaps (CDS) against triple-A tranches of ABSCDOs, and their poor level of diligence in accepting both residential and commercial mortgage exposures. A New York Times article titled “A Lack of Rigor Costs MBIA” (November 13, 2009) reports that MBIA had admitted in legal proceedings that “the due diligence standard for a monoline insurer, which MBIA followed, did not involve looking into the quality of the securities underlying the securities being insured” and that “MBIA could not afford to do much research… if it did such research, it would have to charge much higher premiums.”
Contrary to this public perception, the monoline insurers failed simply because of a deliberate overconcentration in mortgage credit risk relative to levels consistent with prudent risk management and not because of their poor diligence in underwriting and sloppy business practices that MBIA cites. Further, while failures of financial engineering certainly plagued the analyses of ABSCDOs, it was not ABSCDOs or the "Gaussian copula" that brought about the demise of the guarantors. Finally, the large institutions that purchased ABSCDO credit insurance from the financial guarantors could have, should have, and perhaps did understand that the true economic value of these policies at their inception was much closer to zero than to the premiums they paid for them.
While I have a wealth of experience to inform definite views about MBIA's claims of shoddy underwriting standards as they reflect upon the industry, it would be gratuitous (and ungracious) to comment on these in this note. It wasn't the poor quality of underwriting and pricing that led to the above mentioned effective defaults anyway. Instead, through over-concentrating in mortgage credit risk in the chase for outsized and near-in accounting earning streams (i.e. “yields to bonus”) that municipal bond insurance could not provide, monoline insurer executives, with the exception of Assured Guarantee, simply bet their own firms’ houses on the trajectory of nation-wide housing prices and lost. Aggregate levels of credit insurance against individual mortgage backed securities (MBS) reached multiples of guarantors' claims-paying ability, and in some cases, guarantors actually levered their mortgage exposure by investing their own capital (claims-paying ability) in MBS as well. It was the denial of the possibility of systemic economy-wide housing price declines that provided rent-seeking executives cover for this outsized bearing of mortgage credit risk. As long as housing prices would continue to climb NINJA (no income, no job, no assets), fraudulent and other low quality loans would generally prepay and refinance out of difficulty in any event. Risk Management needn't have forecast a crash in housing prices to have advised against such high mortgage exposures, but they would have had to be empowered to report formally – in underwriting committees and to Boards of Directors -- the potential impacts of crash scenarios. Suffice it to say that principal-agent problems precluded the authorization of risk managers to make such formal reporting. Plus, serious consideration of the mere existence of a housing bubble was discouraged (a full discussion of these problems is beyond the scope of this note).
As nationwide housing prices plummeted from their 2006 peaks, MBS (AltA, subprime, Heloc and Hel) began to suffer principal and interest shortfalls and increasing expected shortfalls against which the guarantors were obligated to indemnify or reserve. Recall that the sum total of the MBS principal and interest payments insured was a multiple of the guarantors' claims paying resources (i.e., "capital"). The guarantors had fully exposed themselves to a risk that had offered an arguably low-probability of occurrence before the fact, but ultimately a ruinously high severity given that occurrence. Ironically, all this while holding themselves out as writing insurance to a “zero loss” standard.
In the midyears of the last decade (after the guarantors had built their oversized portfolios of mortgage credit risk), the flow of single MBS insurance deals decelerated to a trickle; a new risk-transfer vehicle, the ABSCDO, absorbed the bulk of the supply of newly-issued credit-risky MBS. (Indeed, the insatiable demand for MBS from the ABSCDO market could only be satisfied ultimately through the referencing of MBS via CDS such that the same exact MBS exposure might appear in multiple ABSCDOs, greatly exacerbating the problem). The monolines shifted the focus of their mortgage insurance businesses to writing guarantees against these pools of MBS that were resecuritized in CDO form.
Thus, despite the monolines well-publicized defaults or "distressed exchanges" on these obligations, I contend that the insurance of ABSCDO tranches is not responsible for the guarantor melt-down; the bulk of these underwritings occurred after the fate of the guarantors was already sealed. By the time that the guarantors started writing significant credit protection against tranches of CDOs based on mortgage backed securities, they had so much exposure to individual credit-risky MBS that their insurance of additional mortgage-backed product was already worthless. This is why they charged so little for writing protection against CDO, CDO-squared, CDO-cubed, etc. (In the above referenced article, MBIA said its premiums were as low as $77,500 for each $100 million of insurance). Virtually any nonzero premium charged for insuring additional mortgage-related exposure whether single MBS or ABSCDOs was above fair value by the time the first ABSCDO insurance deals were struck.
How could the ABSCDO insurance have been worthless -- weren't all the guarantors rated Aaa/AAA when they wrote these “wraps” (guarantor jargon for a guarantee) and CDS? Again, go back to the premise that allowed the guarantors cover for over-concentrating in MBS risk; that is, that the likelihood of a nation-wide housing collapse was insignificantly different from zero. This view permitted a degree of mortgage overexposure that would surely lead to bankruptcy in the perhaps very unlikely event of such an occurrence -- the exact same unlikely occurrence that would lead to claims against the policies written on the AAA-rated tranches of ABSCDOs that were backed by credit-risky mortgage loans. Thus, the purchasers of monoline protection of ABSCDOs were buying mortgage insurance from companies that would run out of resources to indemnify them in precisely the housing market meltdown scenarios in which indemnification would be required! Senior tranches of ABSCDOs could only default (and the guarantors be required to pay claims on their insurance of these CDOs) given a nationwide housing market debacle, which is the exact same scenario that would be catastrophic to the financial guarantors given their existing overconcentration in mortgage risk.
The message bears repeating one more time: "no matter how small or large the probability of a housing market collapse truly was, the ABSCDO protection sold by the guarantors was next to worthless because the insured could not possibly be indemnified when the insurance was 'needed'”. It is important to note that one can argue that incremental municipal bond insurance written on California debt, for example, was likewise virtually worthless, because many financial guarantors had so much California exposure that in the unlikely event that California should default on its debt obligations, financial guarantor capital would be depleted and no insurance client could be fully indemnified1. Risk managers call this "wrong-way counterparty credit risk" and it seems to have reared its ugly head in every crisis since the Russian default event of 1998.
Surely, wrong-way credit risk must have been understood by the sophisticated bankers who have been suffering from the counterparty failures of the monolines in distressed exchanges -- why did they even pay as little as they did for insurance that had no economic value? My guess is that there was another principal-agent problem at play here as well. It appears that many bank CFOs allowed their trading desks to bundle the economically worthless monoline protection they had purchased with the ABSCDO securities they held in their portfolios and then to revalue those securities as if they were suddenly riskless! This revaluation generated instant paper profits and the promise of associated increments to trader bonuses (i.e., "free lunches"). The few basis points the traders paid for the economically worthless monoline guarantees of the ABSCDO tranches was likely viewed as profit-sharing and a cheap means to an instant and much greater upside. In retrospect, it is clear that had proper FAS157 counterparty credit valuation accounting been applied, the bankers would not have exposed themselves to these losses. Some have argued, however, that the bankers who purchased this worthless protection for their ABSCDO exposures could not have possibly understood the counterparty risk presented by the guarantors (beyond the information provided by the AAA/Aaa rating agency letter grades) because in order to do so, they would have had to have pierced corporate veils to uncover the guarantor’s counterparties as well, implying a very complicated information network. Contrary to this argument is the fact that at least one of the prominent monolines published details of all of its MBS guarantees on its website. Had the bankers negotiated collateral agreements with the guarantors such that the latter would have had to backstop the protection they wrote with pledged assets as the credit quality of ABSCDOs deteriorated, the bankers likewise would not now be suffering distressed exchanges. However, collateral agreements were rarely negotiated in the financial guarantee industry, and the guarantors worked proactively to keep costly collateral posting agreements out of financial guarantee contracts.
The lessons of the demise of the guarantors are important, and are rather more basic than arguments of the analytical intractability of complex structured products such as CDOs and MBS. The guarantors built portfolios that were over-concentrated in mortgage risk (both investment portfolios and contingent insurance liability portfolios) in order to chase earnings growth targets that could probably not have been achieved by their traditional business of underwriting municipal credit risk (or if earnings were adjusted properly for the cost of risk). These firms are now in runoff with little to no prospect of ever writing new business. Their clients relaxed prudent standards of counterparty credit risk management (including simple collateral requirements and accounting adjustments for the valuation of counterparty credit) in order to inflate P&L and bonuses and are now realizing recoveries on the order of 10% from the anticipated default (or "distressed exchange") of monoline ABSCDO insurance. By enforcing a respect for simple principles of finance such as "don't put all your eggs in one basket" and "there is no such thing as a free lunch" and constraining the rent-seeking behaviors of their executives, the corporate boards of the financial guarantors and their clients could have avoided the unpleasant consequences of the monoline meltdown.
1The guarantee of California Tax Allocation Bonds (TABS) illustrates an interesting intersection of two potentially fatal risk over-concentrations. While the California regional economy is surely suffering a severe recession, the guarantors are facing heightened probability of payment default in the TABS they have insured, because those bonds rely on revenues linked to assessed housing values to cover debt service.
March 1, 2010