AMBAC: This is not going to work

I have completed a deep dive of AMBAC's insured portfolio. The conclusion: I don't see how they maintain a AAA rating without raising new capital.

The Challenge
First let's consider what AMBAC needs to do to retain a AAA rating. All three of the major ratings agencies have a similar methodology for bond insurers. They need to survive a Great Depression-type scenario. The agencies then estimate how much capital an insurer would need to survive such a scenario. As long as the agencies has capital above this minimum, they get their rating.

Currently AMBAC has between $1.1 billion and $1.9 billion in "excess" capital over what's needed to retain a AAA/Aaa rating, depending on the agency.

What Matters
The ratings agencies have said that mark-to-market losses are "not predictive of future claims" and therefore not a focus of their analysis. I understand where they're coming from, I wrote about a similar idea in AI's recent discussion of Freddie Mac. I'm not sure that alternative methods are likely to come up with more predictive results, but I'm not the one who gets to make up the ratings.

So right or wrong, the ratings agencies are going to focus on a forecast of credit losses when deciding capital adequacy.

Nature of ABS/CDO Insurance
Insurance policies written on ABS and CDOs are in the form of "pay as you go" CDS. What that means is that in the event of a default, AMBAC would only be responsible for paying any interest shortfall and ultimately, any principal shortfall.

So for example, say AMBAC insured a senior, AAA-rated subprime RMBS tranche. Let's say that losses in the pool are such that the junior tranches get wiped out, but the senior tranche only suffers a 20bps/year interest shortfall. AMBAC would only be responsible for paying that 20bps. And those payments would occur over time. This is in contrast to typical CDS contracts, where the seller of protection must buy the reference item from the buyer of protection upon default.

This allows an insurer to absorb credit losses over time. Even if, say, a $1 billion ABS tranche were to suffer a 100% interest shortfall, AMBAC would only pay out annually the interest that went unpaid, probably something like 6%, or $60 million.

Note that this is a good reason why mark-to-market losses aren't everything for bond insurers. Given a default event, the insurer might write down their position entirely, but pay out the claim over an extended period of time. So in terms of capital adequacy, the insurer might be able to earn enough premiums over time to offset losses.

How Bad Will Subprime Defaults Be?
In order to make loss calculations, I needed to estimate what percentage of subprime loans will be foreclosed upon. I think in the 2006 and 2007 vintage, 25% is a good starting point. That is about the percentage of stated income loans underwritten during this period. Of course, not 100% of the liar loans will default, but you got to think the overwhelming majority will.

For 2004 and earlier, I assumed foreclosures would be around 9%, which is the highest level we hit during the 2001 recession. Given that job growth is still positive, I think that's a conservative figure. For 2005, I assumed around 15% defaults.

Recovery should be lower than historical average as well, due to weak home price appreciation. Older deals, like 2003 and earlier might recover at normal levels, but then again, anyone with strong HPA probably will be able to refi or at least work out a loan mod. The overwhelming majority of losses will come from the 2006 and 2007 vintage.

How Much In Losses?
Regardless, I think the way to attack AMBAC's capital adequacy is to consider how much in principal losses their ABS and CDO portfolios are likely to eventually suffer. I believe that if this amount is in excess of the $1.1-1.9 billion figure used by the ratings agencies, then eventually AMBAC would be forced to raise captial in order to retain their rating.

AMBAC's biggest problems will be in their CDO portfolio. I estimate they will suffer $4 billion in losses from their CDO portfolio. This will be almost all in the ABS CDO and CDO of CDO portfolios, which will suffer from the structured squared problem. Losses in other types of CDOs would seem to be within typical historical levels.

Losses in direct RMBS positions look to be in the $2 billion area. Many of their positions will probably suffer no losses at all, as AMBAC usually has significant subordination. But most look like they will suffer some losses.

How Much in Capital?
So how much in capital would they need to retain their rating? Probably at least $2 billion. They have about $1 billion in either loss provisions or mark-to-market losses they've already realized. They should have earnings of around $800 million/year. If we figure that the $6 billion in losses is spread over 3 years, that's about $3.5 billion in internally generated capital. Plus they have about $1 billion in "excess capital" over what they need to retain their rating. That leaves us $1.5 billion short.

You'd assume that AMBAC would want to raise more capital than the bare minimum, so I'm figuring $2-3 billion.

Could They Raise It?
Whether they can raise this kind of capital or not is difficult to see. It will be a question of whether a well-capitalized partner sees long-term value in their lucrative municipal insurance franchise in excess of the losses expected in ABS. I don't doubt that many potential partners would be interested in the municipal business. Munis never default, so writing insurance on them is like printing money. Berkshire Hathaway has expressed interest in the muni insurance business. I'm sure that if Berkshire put, say, $1 billion into AMBAC, then AMBAC could subsequently do a couple preferred offerings. They'd be expensive, but with Warren Buffett already on board, I think they could get it sold.

On the other hand, stronger players in the municipal insurance business may also be looking for capital, most notably MBIA. Even assuming someone like Buffett would consider investing in a bond insurer, maybe he'd prefer MBIA, who has less ABS exposure as a percentage of total par insured. We'll see.

Timing of a Downgrade
Since the ratings agencies are focused on expected losses, it may be that AMBAC and others have a fair amount of time to find more capital. This might allow stronger players (MBIA) to wait for a better market and do a simple preferred offering. Weaker players will likely be forced to raise capital privately. FGIC is particularly in trouble, as their primary owners are two private equity firms and PMI Group. The latter is obviously unable to provide capital at the moment, and the private equity firms (Cypress and Blackstone) don't want to be in a "good money after bad" position.

Consequences of a Downgrade
A downgrade of any of the big 4 insurers (MBIA, FSA, FGIC, and AMBAC) would send chills through the municipal bond market. The result might hurt all of them, even the one not downgraded. Muni buyers may permanently devalue insurance, causing more deals to come uninsured.

As for the downgraded firm, I suspect they'd wind up running off their existing policies, rather than trying to remain a going concern. Perhaps they'd sell to another insurer at a steep discount to book value. Once downgraded, even to just AA, their business model would be destroyed. Ironically, Radian would be in a much better position if downgraded, as their business model was never predicated on any particular rating.

Why Didn't You Say So Before?
Admittedly, I've been more sanguine about the muni insurers until now. My mistake was being overly focused on survival, as opposed to just maintaining the rating. AMBAC would be able to survive the $6 billion in losses if they occurring over time. But I don't think they will keep the AAA-rating without help.

How to Play a Downgrade?
This is a tough call. Going long CDS is a tough call. First, its awfully expensive, as CDS spreads are extremely wide. Second, AMBAC could get a capital infusion or do a distressed merger, and the CDS wind up not paying off.

Its also possible that AMBAC themselves did better than average credit work. In other words, that their insured deals wind up performing better than the average RMBS and/or CDO deal. If in the CDO world, their deals do only 5% better, losses would drop considerably.

Short the stock seems like a better play, but the stock is already trading at less than half of nominal book value. Still, the headlines are likely to remain bad, and any capital infusion is likely to be dilutive. I'd at least avoid the stock.

Disclosure: No positions in any bond insurer, although I own many insured municipal bonds.