Some more thoughts on the MLEC... I still like SivieMae, but whatever.
According to Bear Stearns, who is the biggest lawyer on Wall Street BTW, the MLEC may take residential ABS, just not subprime ABS. They also claim that SIV assets break down thusly:
- 43% financial institution debt (I'm guessing mostly TruPS)
- 23% RMBS (~2% subprime)
- 11% CDO (claiming only 1% is ABS CDOs, so bet that the rest are CLOs)
- Remainder primarily various non-resi ABS.
TruPS is short for "Trust Preferred Stock." It's a kind of hybrid between preferred stock and debt (hence why they sometimes called "hybrids") that financial institutions use to increase their Tier 1 capital. But don't worry about all that, suffice to say that a TRUP is nothing more than subordinate bank/brokerage/insurance company debt.
So let's do a little math. Let's assume that media reports that the MLEC will only buy "highly rated" instruments translates to A and above. Note that I've heard some reports that it will be Aa and above, but we'll be conservative. So if we assume that global SIVs bought their paper at the beginning of the year, what kind of losses are we looking at?
TRUP paper spreads, like any corporate bond, depends greatly on who the issuer is. TRUPS are issued by large and small banks alike, with much of the smaller bank paper going into CDOs. According to Bear Stearns, A-rated CDOs of TRUPS are only 40bps wider YTD. Looking at some A-rated publicly traded TRUP issues, I think that's a little conservative but close. TRUP paper tends to be long-term (similar to preferred stock) so 40-70bps of widening is probably 3-7% in losses. Its possible that non-publicly traded paper would be weaker, but given that CDO spreads haven't moved much I'd say the 3-7% figure is about right.
We'll come back to RMBS paper...
A-rated CLO paper is about 200bps wider YTD according to various sources. AA-rated paper is about +150. Assuming a 8-year average life on the CLO paper, the loss there would be around 15%.
According to Merrill Lynch, auto loan, FFELP student loan and credit card ABS are all about 30bps wider YTD. On assets with 3-5 year average lives, that's a loss of around 1%. Private student loan paper is about 60bps wider, so that'd be a 2% loss, but that isn't that big a market.
RMBS paper is anybody's guess, because every piece is going to trade very differently. Looking at a couple so-called "generic A-rated home equity" indices I get widening of around +900. I can't take at face value Bear's assumption that only 2% of this paper is "subprime" because there is no single definition of subprime. For my money, in today's world, any low-doc loan is subprime, regardless of FICO. Anyway, we're looking at like 35-40% losses on this paper, if you believe the "generic" spread move. AA-paper is drastically better, probably having widened around 300bps, for losses in the 10-15% area.
OK so if we use Bear's percentage figures...
- 43% financial institution paper with 5% losses.
- 23% RMBS with 35% loss.
- 11% CDO with 15% loss.
- 23% Other (mostly ABS) with 1% loss.
That averages out to 12% in total losses. A bit higher than I guessed off the top of my head in my previous post, but then again I was assuming no RMBS paper would be allowed in the MLEC. If we ignored the RMBS, the remaining paper has a 5% average loss.
So now we wonder what kind of paper might be put into the MLEC. I think there are conflicting possibilities.
First you have to consider the Market for Lemons. This is the title of a 1970 paper by George Akerlof, who later won the Nobel Prize for similar work. For any students or young traders reading, this concept advanced is one of the most important concepts you can learn as a bond trader. Basically Akerlof's reasoning was as follows. In the used car market, the seller of the used car, presumably the current driver of the car, has much greater knowledge of the car's condition than any potential buyer of the car. If the car is a lemon, the seller will be particularly motivated to sell it. The buyer will be aware of this incentive, and put a high probability on the car being a lemon.
So the buyer's bid for the car will reflect the probability that the car is indeed a lemon. If the car is not a lemon, the seller will likely be unwilling to sell the car at a "distressed" price. Only if the car is a lemon will a trade occur.
The parallel to the bond market is obvious, particularly when we're talking about securitized products where limited information about the underlying loans is available. If we use the Lemon Theory of Pricing, and we assume the MLEC is only going to buy bonds at distressed levels, then SIVs will only sell distressed bonds into the MLEC. We'll have an adversely selected pool of assets backing MLEC's CP issuance.
On the other hand, we have to examine the roll of CP investors as well as non-SIV sponsoring banks participating in the MLEC. First, CP investors will likely examine the pool of assets backing the MLEC carefully. If the MLEC is coy about what assets are in there, I expect CP investors to assume the worst, and then MLEC's ABCP won't be any better than ABCP is currently. Maybe MLEC can get away with pointing to the guarantee by various banks but it isn't like those banks won't have something to say about the assets as well.
The guarantor banks, notably Bank of America and J.P. Morgan, are going to get a fee to insure the repayment of the ABCP, but neither was involved in the SIV market. As with all insurance policies, the idea is to never pay out. Especially since BofA and JPM aren't getting any side benefit from stability in the SIV world. So I doubt those banks would allow for weaker assets to be put into the MLEC, unless the MLEC acquired the assets at very attractive levels. That would turn the MLEC into a de facto vulture fund, which many have suggested will be the case.
So while we still don't know enough to draw conclusions, considering the motivations of the involved parties, I think we are getting closer. I think there are two broad possibilities:
First, the MLEC only buys very high quality assets, and avoids residential securities entirely. The SIVs will benefit because the sale of assets to the MLEC will give them a nice infusion of cash. The losses they incur as part of the sale will be marginally less than had they gone into the market themselves, but in real economic terms, this will be all but offset by the fees paid to create and insure the MLEC. If the assets are all viewed as very high quality, then I suspect MLEC won't have any trouble getting funding from CP issuers, and every one will hail its creation as a smashing success. However, the whole thing will be nothing other than a tool to obfuscate the balance sheets of SIV sponsors. The real economic benefit will be small, if any. Peripheral as Flow5 called it.
Second possibility is that, the MLEC becomes a vulture fund, with mostly weaker assets sold into it. In this case, the bank-owners will have to pledge a strong guarantee in order to lure investors to buy their CP. As long as CP investors view the MLEC as fully backed by a strong bank like Bank of America, there won't be a problem getting funding. But any kind of semi-backing or "moral obligation" won't fly. The SIVs will sell distressed assets at distressed prices, and the fee paid to the guarantors will be much larger. MLEC's owners will benefit from both the large fee as well as the large carry from these distressed securities. I don't know how the GAAP accounting of this scenario would work, but economically the SIVs selling assets would realize large losses. The broad economic benefit will be somewhat greater, because it might speed up the reemergence of non-agency mortgage underwriting, but I'd still call it peripheral.
In all, I'm still not convinced this whole plan has any real meaning to it. We'll see.