They survived without a government bailout.

I don’t believe that Merrill’s marks on their mezz CDOs should be viewed as overdone by some short term market phenomenon and neither apparently does John Thain. From the Merrill conference call:

———————

“Prashant Bhatia, Citigroup – Analyst

And then, can you give any sensitivity to the [subprime] cumulative loss assumptions, I think, you said 16 to 21%. It’s clearly not linear so if it ends up being higher or lower can you give any kind of sensitivity there on the impact?

John Thain, Merrill Lynch & Company, Inc. – Chairman, CEO

The problem is, as you know, that the, it’s very dependent on the deal structure on a name by name basis. It’s dependent on the vintage of the mortgages. The 16 to 21% was intended to give you the kind of the average range that we’re looking at. It’s very hard to get much more detailed to that other than, unless you actually run the individual positions….

Prashant Bhatia, Citigroup – Analyst

And maybe asking it another way, you’ve taken over $20 billion in writedowns over the last couple of quarters. What percentage of that would you say is a writedown to reflect a market environment that exists today that may not exist tomorrow? For example, how much do you think is really a liquidity type mark based on the environment?…..

John Thain, Merrill Lynch & Company, Inc. – Chairman, CEO

Yes, I’m not sure if you’re trying to get at whether there’s the ability to recover some of the losses.

Prashant Bhatia, Citigroup – Analyst

That’s exactly what I’m trying to get at.

John Thain, Merrill Lynch & Company, Inc. – Chairman, CEO

That’s what I thought. I think on the CDOs specifically, I think it’s not likely that these things are going recover because the fundamental assumptions as to home price declines and cumulative losses, I think, as I said, I think we’re being conservative but I don’t think that we’re likely to get much back on these things.

So I don’t think it’s like just where you have an illiquid market where the liquidity comes back and you recover a lot of it. I don’t think that’s going to be the case here.”

——————

Moreover, the dubious nature of Ambac’s “models” was exposed in their most recent conference call. This is a long winded CYA type of explanation as to why their prior forecast of impairments have been so far off the mark – the plain unvarnished version is ” we finally got around to reading the documents and understanding the ratings triggers, and guess what, a bunch these CDO squareds are going to have their cash flows cut off since the collateral has been downgraded”. I think the credibility of the monolines loss estimates, to the extent it still existed before, went out the window with this conference call.

———————–

David Wallis Sr. MD Risk Mgmt:

“I’d like to focus on the key question — how have we computed Ambac’s likely $1.1 billion CDO impairment, primarily booked in respect to Ambac’s CDO-squared transactions? And what has changed from prior periods?

To cut to the chase, different loss estimates can arise from different methodologies and different assumptions within the same methodology. And what I will be discussing in some detail is an evolution of our analytic focus as we believed events have dictated…

Having talked about the ways in which the CDO-squared loss estimates may be computed, I’d like to make a few comments about their respective results. Leaving aside the market value approach, our analysis produces the result that it is the latter structure approach which produces the highest estimate of loss under our own assumptions. The approach also provides the most transparent and credible framework, given the combination of the current environment and our understanding of the transactions themselves. As stated previously, it is this methodology which forms the basis for the impairment noted.

To illustrate this further, let’s briefly discuss some comparative results and point out what we see as the main flaws in some of the discarded methodologies — the CDO model approach.

Aside from the possible tardiness or inadequacy of ratings as indicators of the probability of default — no criticism is intended here — the model approach has the deficiency of not being able to capture the legal and structural mechanics of the transaction as discussed earlier. As stated, the reason is that at a certain point the dominant driver becomes the affect of the degraded ratings in combination with the legal structure as against the model implied increase in default probability driven by collateral downgrade itself.

Essentially, collateral ratings may give off falsely optimistic signals given the legal structure. The models seize and assigns cash flow to assets that are structurally blocked. Fundamentally, past a certain point of rating degradation, structure likely triumphs over collateral default probabilities, as implied by ratings.

What about the cumulative loss approach? It’s really the same story as before. Here, structure triumphs over cash flow. Potential value in the underlying RMBS, which cusip level drill-down analysis may assert does exist given the vintage distributions, et cetera, simply doesn’t get to be realized by the insured CDO because of the impact of the intervening ratings and structural features. Again, cash flows are blocked.

So to summarize this discussion — and this will hopefully be clear from the above — what has changed and therefore led to the deterioration in loss estimates that we have announced? Essentially two things in combination. Firstly, an exceedingly rapid and substantial set of rating agency downgrades starting towards the middle of October and continuing thereafter.

Secondly, resulting from the above, an evolution in our analytic focus from model or drill down analysis to an examination which concluded that at a certain switch point, legal and structural mechanics will likely triumph over model expectations and the fundamental MBS cash flow analysis.”

————————

Full disclosure: I’m short the stock. But the bottom line is these guys went way off the reservation in deciding to insure structured finance assets for which there were no reliable loss estimates. The losses are going to be enormous and dwarf the pathetically thin capital bases reserved against these risks.

Yes, they won’t go out of business tomorrow, but that doesn’t matter to the rating agencies – it is the adequacy of the capital levels that count – and the rating agencies, at a time and place they view as politically appropriate, will come to the conclusion that the monolines need far more capital to retain a AAA rating. I see no way they can raise sufficient capital without effectively diluting their existing shareholders out of their positions.

Agreed with your points re the merit of considering other research and conference calls. However, not being an institutional investor, I have to rely on the kindness of readers in either passing it on or highlighting where to find it, on those comparatively rare occasions when document are made available in toto.

Your observation about Merrill having junkier CDOs by virtue of having unsold recent inventory is valid; not so certain re Citi, since their exposures came by virtue of their having structured CDOs where the senior debt was commercial paper, and they had a obligation to fund if the CP couldn’t be rolled. My impression is that they had been doing that longer (perhaps as far back as 2005).

Also, remember as of 3Q, the new FASB provisions regarding Level 3 assets were not in effect. Merrill and Citi may well have used indices for valuation purposes, but it would have been kosher for them to use other methods. I haven’t listened to their conference calls, but my recollection (and apologies for not checking) is the writedowns were lower (meaning more favorable to the bank) than the level at which the ABX for AAA instruments was trading.

The media has also blurred over the differences between the two. Ambac is far more heavily exposed to CDOs; MBIA has meaningful exposures to commercial real estate and below investment grade securities, and, as Pershing discussed at considerable length, its problemmatic relationship with ChannelRe.

I don’t see how the holding companies can make it, nor do I see how the insurance subs won’t be downgraded. To me, that is pretty devastating for them. it would be better if I were wrong, but it still seems that a likely endgame is that they are put in runoff mode by the regulators. But if that happens, presumably there will be more forensic work on exactly what they own, and a much better reading on how undercapitalized they are.

Anon of 4:02 PM,

Thanks for the catch. Have fixed it. As regular readers may know, I have a real problem with typos (I have spelled Buffett correctly in other posts). I do appreciate it when people point them out. They are embarrassing!

I did read the document and I very much liked the part on how the bond insurers mark to model. I agree that the model methodology understates the insurers’ loss exposure relative to actual market spread levels, but that is if actual spread levels correctly reflect the level of ultimate losses. So again you get into the issue of predicting losses based on market levels.

As a second point, although retained investment bank CDO exposure is largely super senior AAA (I believe Merrill also specified this in its 3Q call commentary), there are huge differences in CDOs such that you cannot group all super senior ABS CDOs together and state that if one portfolio of super senior ABS CDOs has this percentage of losses, then another should as well. Important distinctions include CDO vintage, high grade versus mezzanine CDOs, asset class collateral composition, exposure to ‘inner’ CDOs, and loss subordination protection levels. These distinctions make HUGE differences in the amount of losses that may ultimately be incurred, especially at the super senior AAA level of a HG ABS CDO. These are complex instruments of which there was little standardization, so it is very difficult to approximate the losses of one portfolio of CDOs based on another without much more information.

Here are some of the commonly attributed reasons for why the CDO portfolios of Merrill and Citi may be worse than that of the bond insurers. The investment banks utilized bond insurers to get some super senior risk off their balance sheets, and fully retained other risk. The writedowns were on the fully retained portion, not the insured portion (yet). CDOs retained by the IBs may have been those that they could not find attractive insurance for, as some bond insurers stepped back from the market in 2007. Also, bond insurers are (believe it or not) selective with what they insure, so that which was not insured could be deduced to be riskier and therefore less attractive to the bond insurance underwriter.

Furthermore, retained IB exposure (again, refer back to Merrill’s 3Q conference call commentary) of CDOs is primarily from this year; much of the older vintage CDOs are no longer on the books. CDOs from this year contain collateral from early this year and 2006, which are the worst vintages. CDOs containing older vintages are less toxic. While bond insurers certainly have plenty of exposure to 2006 and 2007 vintage CDOs, the proportion of their insured CDOs in the worst vintages is likely less than that of the retained portfolios of the IBs.

Thanks for your post, your commentary and the research you have done. I am not attempting to be combative, I am simply raising points that I know to be valid based on work I have done all year. I’m not suggesting that Pershing’s analysis is entirely invalid, but I do think that their arguments (and those of a lot of bears) had some problems in them. This is unfortunate because it became relatively easy for analysts such as myself and those in the Street to refute those arguments. Now, as the downgrades have begun, we are forced to admit that Ackman was right in his bearish call. But it is still relevant to bring up issues in methodology that may be incorrect, because the story is far from over. As the focus now turns to runoff scenarios, there will continue to be speculation and uncertainty regarding the actual levels of losses incurred by the insured exposures.

I would suggest to you that you not place too much focus on a single document. If you have access, I suggest the UBS piece put out on January 8th, which is the most comprehensive piece put out recently, as well as the reports from Citi Equity and BofA Equity this week (in which they threw in the towel on their Buys). Beyond simply gloating about the fact that these guys were wrong, you should pick up some of the other issues at hand and focus on what may come going forward. If you are truly interested in this issue, I would also suggest that you look closer at these companies, examine their disclosure (ABK’s is the best), listen to their past conference calls and presentations, and look at some of the Street research from the past several months so that you can understand why, from an analytical perspective, many did not believe prospects for these companies would become as dire as they have in the last week.

Did you read either what I wrote, or what the Pershing document said? It most certainly did NOT base the CDO loss estimates on the indices. It used the average of Merrill and Citi 3Q writedown. Both firms claimed their holdings were almost enirely AAA (Merrill said AAA, Citi said “super senior”). This would make their exposures similar to the sort that the monolines are exposed to.

Pershing has concluded that MBIA’s holding company will run out of cash in a best case scenario, by the end of 2008, but under a different scenario, in six months. I suggest you read their reasoning before taking issue with it.

The only place Pershing used indicies as a proxy for loss exposure was with CMBS, not CDOs.

Your comment indicates you have not read the document. I suggest you do so.

First, the CDO loss projections are simple approximations from the market prices of indices such as ABX and TABX. These indices are largely hedging vehicles that have been driven down disproportionately by massive Street hedging, as well as momentum-driven fast money bets. Many analysts concede that the level of cumulative losses implied by the ABX is higher than their projections. The indices are also a poor proxy for the broad market, as they are weighted toward lower quality deals.

Furthermore, the quality of bond insurers’ portfolios is higher than that of the broad market. You can verify this not only by checking out each insurer’s disclosure of the vintage and asset class composition of its CDOs, but also by referring to S&P’s December report that stated that the downward ratings migration of the broad universe of RMBS and ABS CDO securities was over 4x that of similar exposures insured by the monolines.

There is a lot of debate and uncertainty on how to measure projected CDO losses, so you could argue my points above. However, regarding CMBS, almost every CMBS analyst would agree that extrapolating cumulative losses of CMBS from the current level of CMBX is ridiculous. The extent to which CRE will fall is debatable, but current CMBX levels are driven by contagion from other stressed structured products, namely ABX.

Also, the argument against the mark to model approach that market spreads are a “probabilistic assessment of default risk and recovery” ignores the fact that current CDO and RMBS spreads incorporate huge liquidity discounts, in that any sellers are forced sellers at distressed prices.

The Pershing document linked above, if it is the one you’re referring to (sounds like it is), is from the end of November. Obviously Ackman and co. have gotten the call right, but the monolines’ downfall since November has been almost entirely due to market issues, rather than the fundamental risk of their insured portfolios going up.

The decrease in confidence in the monolines because of the efforts of those such as Pershing has driven down the stock price of the insurers, which in turn reduces their financial flexibility and their capital adequacy. This has continued to spiral into a self-fulfilling prophecy in which the bears may turn out to be right just because the market believed they were right.

The actual level of losses incurred by these companies may not be known for many, many years, as the obligations are relatively long-term in nature and take a while to play out. In particular, since the insurers hold the super-senior positions on the CDOs they insure, if and when insured CDOs breach triggers they will choose not to liquidate but to accelerate payments to their senior tranches, thus reducing their par at risk.

Furthermore, even all loss subordination is breached and the insurer has to pay a claim, the insurer is only liable for scheduled interest and principal payments when due, not the entire amount all at once.

These are all important points to consider when you make broad, sweeping conclusions like “there is no way these companies will survive.”

Forget about highway robbery this is far more profitable.