Ambac Financial Group Inc (AMBC) 2007 Q3 法說會逐字稿

At this time, all participants are in a listen-only mode. A brief question-and-answer session will follow the formal presentation. (OPERATOR INSTRUCTIONS). As a reminder, this conference is being recorded.

It is now my pleasure to introduce your host, Mr. Sean Leonard, Chief Financial Officer of Ambac Financial Group. Thank you, Mr. Leonard, you may begin.

Thank you. Welcome everyone to Ambac's third quarter conference call. I'm Sean Leonard, Chief Financial Officer of Ambac. With me today are Robert Eisman, Controller; Peter Poillon, Investor Relations; and David Wallis, Senior Managing Director responsible for Portfolio Risk Management. David will be available to answer any credit-related questions at the end of my prepared remarks on the third quarter earnings.

Our earnings press release, quarterly operating supplement and a short slide presentation that summarize the quarter's results are available on our website. Also note that this call is being broadcast on the Internet.

Fourth quarter 2007 earnings will be released on January 30, 2008 at 6:00 a.m. with a conference call later that morning. During this conference call, we may make statements that would be regarded as forward-looking statements. These statements are based on management's current expectations. I refer you to our press release for factors that could change actual results.

At the outset, let me say that looking beyond the marked to market adjustment that we announced a couple of weeks ago and about which I will speak later, Ambac had a good quarter with solid new business production. Overall, the business environment has improved after several years of narrow spreads and pressured pricing. We saw solid demand for our core financial guarantee product.

In this quarter, we recorded a net loss of $360.6 million or $3.51 per diluted share compared to net income per diluted share of $1.98 in the third quarter of 2006. The current quarter was impacted by the previously announced negative mark-to-market adjustment amounting to $743 million pretax, related to our credit derivatives portfolio.

While Ambac reports net income in accordance with generally accepted accounting principles or GAAP, research analysts make certain adjustments to net income to calculate the reported estimates to arrive at an operating income number that reflects how they view the underlying performance of our business. Therefore, to enhance investor's understanding of our financial results, we continue to provide information on the items that analysts adjust out of GAAP net income to arrive at their current estimates.

Those items are as follows. First item -- net aftertax gains and losses from investments, securities and mark-to-market gains and losses on credit, total return and non-trading derivative contracts. In the third quarter of 2007, Ambac recorded net aftertax losses amounting to $553.5 million or $5.39 per diluted share, and that net amount is added back to our GAAP results. Again, the over-long majority of this loss relates to mark-to-market loss from credit derivatives that I will discuss later. This quarter's results compares to third quarter of 2006 when we recorded net aftertax gains of $6.5 million or $0.06 per diluted share that were backed out of our GAAP results. On this operating basis, earnings per diluted share were $1.88; down 2% from the comparable prior period.

Second adjustment. Some analysts also back out the aftertax effect of accelerated premiums earned on obligations that had been refunded and other accelerated premiums. Total aftertax accelerated premiums amounted to $9.6 million or $0.10 per diluted share in the third quarter of '07, which compares to $14.5 million or $0.13 per diluted share in the third quarter of '06. On this basis, our core earnings of $1.78 per diluted share were basically flat to the comparable prior period of $1.79.

Now turning to credit enhancement production, or CEP. CEP, which represents gross upfront premiums plus the present value of estimated installment premiums on insurance policies and structured credit derivatives issued or assumed in the period, came in at $431.1 million, up 99% from $216.2 million in the third quarter of 2006.

Now that me take you through some of the details of our production by sector and how each has been affected by current market conditions. First off, public finance. Public finance CEP was $151.3 million, up 68% from the third quarter of '06. Overall market issuance of approximately $93 billion was up 13% from prior year while total market penetration, which is the percentage of bonds issued during that period with financial guarantee insurance, was approximately 50%; flat compared to third quarter of '06. In the current quarter, we benefited from the closing of a large tax-back transaction a quarter ago. Overall, the public finance markets were little affected by the turmoil experienced in the structured finance markets; the market characterized by strong issuance and strong competition amongst the insurers.

Next, structured finance. Structured finance CEP was $108.2 million, up 38% from the third quarter of '06. The volumes in many asset classes were significantly down during the quarter due to the much publicized turmoil in the markets. CEP in the third quarter was driven by a strong pipeline of deals from earlier in the year along with some very attractive transactions that were underwritten and closed within the quarter. For example, we underwrote a large student loan transaction for an issuer that had over the past several years executed transactions without insurance.

Spreads are demonstrably wider in most structured markets and pricing has moved up accordingly. As volumes continue to improve over the next several quarters, we expect the impact to be positive on new business production.

During the quarter, our strongest writings were in student loans, investor utilities, offset by lower writings in CDOs and auto securitizations. During the quarter, we underwrote $2.6 billion of direct RMBS transactions, all rated AAA, and eight new CDOs of corporate assets, also all rated AAA.

International CEP in the third quarter came in at $171.6 million, up 259% from the third quarter of 2006. Strong writings across several asset classes and geographies were included in the quarter, the most significant of which was the New York Tunnel refinancing. Again, this quarter's production is a reminder of the lumpy nature of this segment's production, as 77% of CEP came from large transactions. Nevertheless, it remains a good source of profitable growth.

Turning to premiums earned. Net premiums and other credit enhancement fees earned, excluding refundings, increased to $198.4 million, up 4% from the third quarter of 2006. Public finance earned premium excluding accelerations was flat as public finance earnings have been impacted by competitive pricing, the mix of business in recent years, and the high level of refundings in the book recently, especially over the first six months of this year.

Structured finance earned premiums grew 8%. Excellent recent production with strong pricing and asset classes such as commercial ABS and CDOs has offset the negative growth trends caused by the high level of pre-payment this sector saw in recent years and even earlier this year. International earned premium increased 3%, impacted by earnings from the strong production generated over the past year and this quarter.

Accelerated earnings from refundings and terminations amounted to $16.4 million pretax in the third quarter of 2007. That is down 33% from the comparable prior quarter as refunding issuance declined significantly during the quarter. Our deferred earnings, representing future earnings on premiums already collected and the future value of installments stands at $6.6 billion. These deferred earnings will be recognized as earned premium and other credit enhancement fees in the future over the life of the related exposures.

Some comment on investment income and our investment portfolio. Investment income was $115.8 million, up 8%, substantially due to growth in the portfolio driven by strong operating cash flows in the financial guarantee business. Our investment portfolio ended the quarter in a net unrealized gain position of $100 million before taxes, down slightly from a $113 million unrealized gain position at June 30, 2007. Our portfolio remains a very high-quality portfolio that contains no subprime mortgage exposure.

Some comments on mark-to-market and unrealized loss from our derivative portfolio. Ambac's portfolio of credit derivatives totaled $71.5 billion at September 30. The credit characteristics of CDX execution and insurance policy execution are similar. For example, in the event of default, claim payments are made over time as principal and interest comes due. However, since CDS transactions may not involve a cash bond or do not require the guaranteed party or beneficiary to hold the bond, they must be mark-to-market under generally accepted accounting principles.

During the third quarter, most structured finance asset classes experienced significant price declines driven primarily by the uncertainty related to the ultimate outcome of residential mortgage losses. These markets saw reduced demand from investors for certain asset classes and forced selling by structured and/or leverage investment vehicles throughout the quarter resulting in a lack of liquidity and ultimately distressed prices. These factors led to our $743 million negative mark-to-market in the quarter.

Ambac observed pricing changes across the entire credit derivative portfolio but witnessed more dramatic declines in our CDO of ABS exposures. Approximately 75% of the mark is derived from our exposures to CDOs of ABS with the remainder attributed to CDOs of corporate assets, both loans and bonds.

We have updated our website disclosure today, providing detail on our CDO of ABS portfolio, including updated Ambac transaction ratings and underlying rating agency collateral ratings. Since all of our credit derivative transactions are performing and are rated internally above investment grade, no adjustment to operating earnings is considered necessary.

Some comments on losses and loss adjustment expenses. Loss provisioning amounted to $19.1 million in the quarter compared to a negative 2.5 million in the third quarter of 2006.

Now let me provide some details on the loss activity during the quarter. Total net losses at September 30, 2007 amounted to $273.8 million, up from $251 million at June 30, 2007. Total loss reserves include case basis reserves, which are recorded for bonds in our portfolio that had defaulted, and Active Credit Reserves were ACR. Case reserves of $107.1 million at September 30 are up $59.8 million from June 30, primarily due to two recent HELOC transactions that are clearly underperforming our original expectations, and as a result, have been internally rated below investment grade.

We have paid very little on net claims to date on these two transactions; approximately $280,000. In one instance, Ambac has a fully funded first loss protection which hasn't been fully utilized, and therefore, has not resulted in a net claim payment. For the entire portfolio, total net claim recoveries received during the quarter amounted to $3.7 million.

ACR of $166.7 million at September 30 are down $37 million from June 30, driven by net favorable credit activity resulting in ACR reserve declines, driven primarily by a combination of upgrades and exposure reductions in the public finance sector, offset by some additional reserves set up for existing RMBS transactions and one RMBS transaction newly rated below investment grade.

Our below investment grade exposures remained flat during the quarter at $4.7 billion; were less than 1% of our total portfolio. The below investment grade mortgage-backed and home equity asset class balance increased $330 million from June 30 to $1.5 billion. This increase was offset by declines in public finance, particularly in the public transportation asset class.

While we are pleased with the overall credit quality of our portfolio, we continue to observe weak credit conditions throughout the U.S. mortgage market that has and may continue to impact our RMBS and CDO of ABS portfolios. Ambac expects these conditions to persist into 2008.

To date, even under these stressful conditions, most of our transactions are performing; remain rated within the investment grade category. It is also noteworthy to mention that Ambac has updated its website disclosures regarding our consumer asset-back portfolio, which includes our entire RMBS portfolio. As always, we will continue to actively monitor these transactions, closely analyzing collateral performance and then consider structural protections available to us.

Next, some comments on operating expenses. Gross financial guarantee underwriting and operating expenses for the third quarter of 2007 amounted to $50.6 million. That's up 11% from $45.7 million in the third quarter of '06. The increase is primarily due to increased premium taxes driven by higher ratings during the period, particularly in jurisdictions with higher than average premium tax rates. In the third quarter '07, Ambac's income tax provision reflects adjustments to our estimated annual effective tax rate. Since we are only nine months into our year, we have recognized only 75% of the income tax benefit related to the unrealized mark-to-market loss on a GAAP basis. We expect, depending, of course, on the final full year's mark-to-market results, that the remainder of the tax benefit will be recognized in the fourth quarter. Third quarter operating earnings do not include the unrealized mark-to-market adjustment and its related tax impact and therefore, will not have a similar adjustment in the fourth quarter.

Next, some comments on the financial services business. Our financial services segment is comprised of the investment agreement business and the derivative products business. Financial services net revenues, excluding realized and unrealized gains and losses, were $9.8 million; down 26% from the comparable prior period, primarily due to lower revenues from each of our financial services products in the third quarter '07.

Next, some comments on return on equity and our capital position. While GAAP ROE was negative due to the unfavorable mark-to-market, our ROE on an operating basis was 12.7%. Our adjusted book value per share amounted to $87.88 compared to $90.63 at June 30, 2007; declined as a result of the net loss reported this period.

Management believes that at the current time, conservation of capital is prudent for several reasons, including new business opportunities and the improving business environment and as a cushion against expected future downgrades of mortgage and our CDO transactions by the rating agencies. As always, we remain committed to managing our capital structure to a level that is consistent with a AAA rated company while maximizing shareholder value.

So in summary, looking beyond the mark-to-market adjustment, we feel good about our operating performance this quarter. We are pleased with our solid new business production, our strong pipeline, and the better overall pricing environment. We expect to see stronger demand and more profitable opportunities for our financial guarantee product into 2008.

That concludes my prepared remarks on the financial results. I would now like to open it up for questions.

(OPERATOR INSTRUCTIONS). Ken Zerbe, Morgan Stanley.

I kind of have two broad category of questions here. First of all, in terms of your mark-to-market loss, I estimate that your mark-to-market is somewhere in the range of, say, 2.5% of your CDO of ABS exposure, roughly speaking. Merrill Lynch came out today and announced its write-downs for the same period on the same Super Senior AAA exposures -- you know, high grade, mezzanine, et cetera -- ranged anywhere from, I think it was 19 percentage points on high grade ABS to something like 57 percentage points on mezzanine CDO [rates].

My question is, why is there such a huge difference between Merrill, who is basically writing off -- or writing down half the value of their Super Senior AAA CDO exposures versus Ambac, who is only taking a couple of percentage point write-down?

Without knowing exactly what's in Merrill Lynch's portfolio, I can't really contrast the two as to exactly what's in that portfolio and how those transactions are structured and what other protections are available within the transactions.

I can comment on our process and what we've done. Our process is, Ambac effectively -- we do a discounted cash flow approach. The way we get our key inputs, which include underlying prices from broker-dealers into that process. So one key assumption is the underlying broker/dealer quotes on the underlying reference obligations. So we would go to the broker-dealers at the end of the quarter and ask for prices. So that is a key assumption that gets into the discounted cash flow model.

Also, the nature of our contracts -- I don't know if they're the same pay-as-you-go type language, but what we do, another key assumption in our model is that we look at the relative change of the pricing for the bonds during the quarter. So effectively, Ambac does not capture 100% of the spread on a bond at the inception of the contract. We're getting paid anywhere from one-third or higher as a percentage of the available spread -- the credit spread on the bond. So that is another key assumption that goes into our model. And then we look at that obviously each time we mark the portfolio.

And we also looked at the weighted average lives of our transactions. And then we discount what that expected premium or the new premium, if you will, that would be charged under that exposure back to a present value, just using appropriate discount rates. So that's the process we utilize. The key input to that and which drives most of the value is the underlying prices that we're obtaining from the banks.

No, understood. I definitely recognize that there's differences probably in terms of process as well as structure of the deals. It's just very hard for me to reconcile that even accounting for the differences in the types of CDOs that a Super Senior AAA high grade CDO can be marked down so much more at one firm versus another. But -- all right.

The other comments I'll make is, of course Merrill came out this morning, I haven't had a chance to really look at that. And if you look at the other financial guarantors and the numbers that have been pre-announced --

Yes, they're very similar to your write-downs.

You know, obviously, ours are -- what we're seeing, as I mentioned, we're seeing higher numbers in the CDO of ABS with the mortgages rather than the corporate. But nonetheless, we've seen some pretty extreme price movements into some of the underlying transactions.

Okay. And then the second category of question I had is, I think American Banker came out and said S&P is likely to downgrade another $20 billion of subprime RMBS in the future following its downgrade a week or two ago. Now I understand that your internal ratings -- and correct me if I'm wrong on any of this -- but your internal ratings on the four CDOs that you downgraded were based on S&P's recent transactions -- recent downgrades.

Is it safe to assume that if S&P comes out with another $20 billion of downgrades, that we are going to see additional internal Ambac downgrades of your high grade and mezzanine portfolio?

Ken, let's see if I can help you on that one. It's David Wallis. Just so we're clear, that the cutoff date in terms of agency ratings for the CDO of ABS portfolio was, I think, last Friday. It encompasses the Moody's data. It didn't encompass the recent S&P data. However, on a forward-looking basis, in the event that the Moody's data didn't downgrade 2007 collateral, be it RMBS or CDO collateral, then we applied a pretty substantial notching of that not-downgraded collateral, if that's clear. And so to the tune of -- I think it was two notches for AA up to -- I believe it was six or 7 notches for BBB. The other point obviously is that it's highly likely I think that the vast preponderance of any additional S&P downgrades have already been, at least somewhat -- and I think one's expectation would be in the majority encapsulated by the prior Moody's downgrades, or indeed, the additional notching that we've done.

So, no, I think that around the edges -- [the deal] that Moody's didn't [have a] downgrade or possibly didn't rate. But I think that we've done a pretty decent job of banging up the ratings that we used to re-rate the transactions that we've re-rated.

In passing, I'll note also that obviously we make a point at our underwritings to notch the underlying ratings that we see. We've talked about this, I think, on prior calls. It's interesting to me to note that I think it was in three or four deals -- obviously they still retain a AAA rating, these deals by us, but in three or four deals, and notwithstanding all the rating downgrades we've seen and the further forward-looking notching prices that I've outlined; in fact, the average rating was higher at -- I beg your pardon -- lower underwriting than it is now. So that points, to some extent, to the pretty extensive adjustments we do make at underwriting.

Great. And then the one final question I had is, if the rating agencies adjust -- because I think the rating agencies right now have all of your CDOs, regular and squared included, at AAA. If the rating agencies were to downgrade the CDO transactions to your internal ratings, could you tell us what the additional capital impact would be on your capital?

Yes, Ken, it's Sean again. I'll give you the number. The overall number including if the -- under Moody's; I'll give you the Moody's statistic -- if you were to re-run the models assuming those lower levels of rating, Ambac's rating, if you will, and also include the mortgage transactions that we also downgraded during the quarter which generated some reserve movements, that entire number is $160 million.

Okay.

I'll just jump in and correct myself. The notching on the BBB deals were six notches. It was two at AA, three at A, six at BBB. So that's the notching in respect to those deals that haven't yet been downgraded by Moody's.

Okay, great. Thank you very much.

Geoff Dunn, KBW.

Sean, can you first update us on the watchlist? How many deals related to your CDO book and RMBS book were on there last quarter? How many are on there now? And what's the migration pattern look like?

Sure, I'll take that, Geoff. Clearly our CDO portfolio is investment grade; very, very large investment grade. So there's no change in respect to all of the extent of watchlist on our CDO portfolio whatsoever.

In relation to the RMBS portfolios, and I'm talking about the direct business here. In fact, all the deals that we've referenced were on various watchlists and obviously have been downgraded in large part prior to today's announcement. So I don't think anything particularly new in terms of what's on the watchlist. I mean, there will be always movements around the edges in terms of some transactions, but nothing really substantial. So we haven't been hit out of the blue by something that we didn't have on some kind of list. I think that it's fair to say that the rapidity of decline in one or two transactions has been surprising; [I'll just] talk about that. But in respect to the major (technical difficulty) watchlist, really there haven't been major changes.

And at what tier of the watchlist do reserve provisionings really pick up? And what's your pool of RMBS right ahead of that break point?

Well, our policy is really to obviously look at each individual deal. We, as you would expect, do not reserve where we have an invest transaction; that doesn't make any sense. So where the focus goes up in a reserve is clearly at the investment grade level. So I'm talking about below BBB minus [BAA three]. It is the case, and it continues, that we have several deals that are rated well, well above that, that in fact might be on our watchlist. So for example, I can think of one RMBS deal that is AAA, and actually as it turns out, is performing just fine; that's been on our list for some time. The particular case in point here was an originator that went bust.

Obviously what we're trying to do is to look forward in relation to our watchlist. It's interesting that we've posted the reserves that we have this quarter as we're going through the claims that we've paid, net receipt overall, obviously, and very small claims in relation to the mortgage-backed deals. So we are trying to look forward in respect of our lists and reserves.

Okay. And then Sean, just a last follow-up on the FAS 133 mark. I think there's still some confusion as to why you don't just purely use the quotes. Just quickly again just highlight the bullet points of why there is a model that the quotes are an input to, rather than just being used purely.

Yes. If we were to use the bond course, when the transactions originated, the underlying cash spread on the bond is going to exceed the premium that's being charged on a particular transaction due to the various tailoring of the contract and the lack of funding and liquidity type issues inherent in the contract.

So for instance, if a particular -- at underwriting or at transaction pricing, we may be charging a premium that is one-third of the originated cash bond spread. So that is used, that particular percentage is used throughout the life of the contract unless we see a reason to change that as a kind of the synthetic price for the risk that we're taking. So if that particular spread would move from 30 to 60, we would move up the price that we would charge -- our theoretical price that we would charge underlying the contract, say, from 10 to 20. And effectively that additional 10 basis points that would be theoretically charged would be discounted over the weighted average life of the transaction to arrive at an unrealized loss amount.

So, you're tracking the actual quotes, but it's on a relative basis and present value [defective]?

Yes. If not -- and this is in some of the new accounting standards, but you need to calibrate the model -- if not, you would have losses upon origination of the contract. And really you're trying to adjust for the differences between the price of an underlying cash bond and the price of the financial guarantee/credit default swap that we're trying to ascribe a value to.

[Robert Hansen, OIN.]

I wanted to get a couple of clarifications on two questions that Ken asked and then ask one of my own.

The first is on the $160 million of incremental capital. Could you just clarify that that is the amount that Ambac would need to raise? You have access capital today; is that just the, in effect, that hit that you would take in further losses if you were to go to the ratings that you talked about?

The second question is -- and why don't I just get these off and then we can get them -- on the -- I didn't look at Merrill's either, but you may want to highlight the difference between how a CDO looks with respect to how it is priced versus the liquidity risk that you actually take when you're backing a CDO. And when we invested in Ambac, we invested in it because of that difference.

The third thing I would like to ask is that you said that 75% of the present value mark-to-market on the unrealized loss is attributable to the ABS of CDO transactions. My question there is, is that how much is actually left to write down? And as you were and if you were to pay claims under the most adverse circumstances, effectively I take it you would pay cash essentially out of the mark-to-market that you've already taken. And it would seem to me that I would want to understand what the remaining risk is in terms of further write-downs and whether that is an accurate way of thinking about the actual incremental exposure going forward on the ABS of CDOs?

Okay. I'll start off with those questions. On the capital, $160 million, what that was to represent is the impact to the levels and it would be a reduction to our access capital levels, which have been most recently reported by the major rating agencies is anywhere between $1.1 billion and upwards towards $2 billion. So that would be a reduction in the access capital levels. It would not (multiple speakers) --

That's not a meaningful number with respect to the overall capital level. In other words, it's $100 million on $1 billion. It doesn't pierce through that excess capital number.

That's correct. We're trying to give folks a sensitivity towards downgrades as it relates to the overall credit modeling and the strength of the AAA, obviously with those levels of access capital. And this result is a percentage of that excess capital level.

Okay. And the second question?

Yes. Regarding the Merrill Lynch, the CDOs -- again, it's very difficult for me to sit here without having a lot of detail as to what underlies that particular portfolio to make meaningful comments. I just can't help with that at this particular point in time. [I would] obviously look at that and may, even after that, may not have comments. But I'd have to look at that to see what the differences would be. I assume it's collateral and structure, but I'll have to take a look at that.

I would guess that it probably may not be a distinction between cash and synthetic, i.e., the liquidity risk. Because clearly what we're doing, as Sean described earlier, is taking the relative change and the relative merits of synthetic or cash are constant. So I think it probably may not be that. It's, I guess, structure and possibly term, et cetera, et cetera.

Okay. That's a good point.

And the third question? Really, with what is left of the ABS of CDOs on a present value basis and so forth?

Yes. That's obviously dependent upon levels of exposure we have, Bob, and where the prices would potentially go from here. So it's hard to look out into the future and predict what may or may not happen. But obviously from on our CDO of ABS, we have -- we've disclosed our total exposures. So that would be obviously the outer bound. And we've given that by individual transaction and we've entered into so folks can get a sense for obviously how we look at the transaction from an internal rating perspective and a variety of other data relating to those individual transactions. So that would be the total exposure. And the exposure by individual transaction, that clearly would be the outer bound. It's hard -- obviously that's just the exposure. The pricing points along the way and what may happen is something that's hard to predict.

I will make the comment that on the corporate side, that might be a little bit quicker to come down than the mortgages because of the uncertainty, obviously, in the mortgage space.

Right. But the point is that you -- mark-to-market, you're going to in effect -- what the market -- what you're saying is, is that you're going to pay cash claims over a period of a number of years. And when you present value, the roughly $2.5 billion, you are imputing a certain cash claim present value at over a period of time. My question is, is that you can't pay more cash claims than a notional present value of what you have outstanding. What do you have left to write down on those?

Well, the notional amount that's outstanding is what we've disclosed on our website, which is the $29 billion; what the mark-to-market unrealized loss is meant to represent. It's not meant to represent a proxy for potential claim payments. We're looking at the underlying credit and describing expectations which manifest themselves in internal ratings. So that gives you a sense for our view of the potential credit and the potential to pay claims on the underlying transaction.

As David pointed out, our methodology, the -- trying to take -- we do take the broker quotes. There's an element of liquidity and market distress and other things like that in the underlying price quotes. That does not translate into expectations for claim payments. But if we were to pay, just to follow-on your point, if there were to be cash payments, obviously, that would come out of that particular mark-to-market liability.

Yes. I was just referring to the specific mezzanine, the ABS of CDOs that have been -- you know, the 24 -- 20 -- you know, 24 through 27, and the proportion of the write-down associated with those.

Let me tell you, the market has been high on those transactions. We don't believe it reflects credit but yes, the market has been higher. So as Sean says, I think on a GAAP basis effectively what would be happening, should we pay claims -- and we do not expect any claims -- then in a sense the moneys would be transferred from one account to the other. We've already taken the net income hit there, obviously.

Gary Ransom, Fox-Pitt, Kelton.

I had a question of the two case reserves. Whether you could give us some metrics as to why those were performing so much worse and warranted the case reserves. I guess what I'm really trying to get at is, what's the differential -- how much of a difference is there between what those two were doing and how they were performing versus the rest of the RMBS that you have where you did not put up any case reserves?

I'm very pleased to say that there's a big difference. Let me -- [I was about] to take the opportunity to talk about the two deals. You'll obviously see them on the website if you haven't already. I guess the first point I'd make is that the -- one of the [head officers said] it was a kind of idiosyncratic deal, candidly, in terms of some of the structural features, which I'll allude to in a second. I think that the deal itself had an unusual degree of seasoning, i.e., the loans at issue of the prospectus, if you will, launch of the deal, were already somewhat seasoned.

The structure of the deal was what's termed a no-OC down deal. So the idea here -- and ideas have changed on this particular point which I'll explain -- was that excess spread should build OC, i.e., build credit enhancement, if you will, in the deal. And this particular transaction, again, there's an idiosyncrasy. There was a holiday, a spread holiday, i.e., the moment that excess spread would be used to build up OC was somewhat deferred.

If you put these idiosyncratic facts together, i.e., you have a seasoned deal, you have a second mean product, obviously, a HELOC deal, and the feature there is that losses tend to come in somewhat earlier than first lien deals, and you put together the absence of starting OC with seasoning with, I have to say, pretty egregious performance of the pool, that's what led to the very early loss.

I will say that the overall characteristics of the pool at face value were not outrageously [paused] as a weighted average FICO 700, a weighted average CLTV 86, very high percentage owner occupied. So a lot of the normal metrics that you'd look at look sensible. However, it is fair to say that performance has been pretty poor and with the structural features I've outlined, this has led to obviously very early claims. So the natural question would be, and obviously you asked it, have we any more of these? And the answer is we haven't.

So we have obviously looked at the portfolio and look for this particular combination of facts and any lessons that we can take away from it. So we have not got the transactions that have zero enhancement and have the [get go]. So either there's OC or there's subs, in addition, obviously, to excess spread. I will say that since the transaction was done, the market has clearly changed. People including ourselves obviously have reacted to what is happening. And now deals just don't get done on a zero down OC basis.

In relation to the other transaction, it's again a HELOC transaction; a very idiosyncratic performance again. A pool that looks broadly sensible, slightly higher FICO, slightly higher weighted CLTV, but again, 91% owner occupied, 32% full doc, and so on and so forth; not significantly seasoned; three month seasoned. That's average for that kind of transaction. Just a very poor performing pool. Obviously, we've noticed this for a while. It's been on various lists. But the performance was such that we decided that we needed to take action on it.

No claims have been paid obviously on that transaction. As Sean mentioned in his discussion at the beginning, there is a fully funded source of enhancement in the deal. And let me just clarify that. This is a source of external enhancement. So it's exogenous to the deal. What it is in fact is a [seal] and a credit linked note, which has been sponsored, if you will, by a hedge fund. There's no counter-party risk; it's fully cash collateralized. And in effect what's happened is, claims have come through, we've paid them -- or paid it, I should say -- and obviously drawn down on the collateralized deposit of the hedge fund which took the deal up into the BBB category.

A feature of both deals, and it's stuff that -- I'm not saying it's a lesson learned but it's something that's interesting and we're alert to is that the deals were bank shelves. So that's not to say -- sorry, investment bank shelves -- i.e., that they were gatherers, if you will, so that an investment bank shelf will buy on a whole loan basis or a bump basis effectively purchase whole loans in various size parcels and slice and dice and securitize. So this isn't what we term bank HELOC paper where obviously it's their paper, they originated it and quite often they'll have the sub piece also.

So that's one of the features that we've looked at in relation to the business that we have been doing. And that is fairly clearly reflected, I think you'll find if you look at the website. In terms of recent HELOCs, for example, that we've done, you'll see that we've done a couple of deals for Wachovia which comprised a vast majority and of the addition to the HELOC portfolio.

So I think in summary, that hopefully gives you an idea of what happened. No, we don't think there are any more reserving. We do think it's -- we think we've taken appropriate reserves. It's an estimate. It's pushed out into time. We are a company that likes to take our lumps, clearly. If we see something we don't like, you can clearly see that in relation to claims paid versus reserves booked. And there are some clear lessons that we've learned. And we'll fully take that into account.

Just a follow-up. This hedge fund relationship -- is this the only relationship, is just this fund?

We started as a very small number. This is a fund that candidly saw a good risk reward and obviously, they were wrong as we were in this particular case -- in a particular tranche of this exposure -- a subordinate tranche to us. So one of the things that we are spending some time on here is looking for new and innovative ways to transfer risk, to enable us to play at the sort of risk reward points that we wish to play. And in this particular case this opportunity became available and we took it.

Just one additional question. On the ACR side, as you look across the credit portfolio, I guess there has been some takedowns on the muni side, are there any other offsetting increases, even though they might be minor, in other sectors?

Well, there's some modest uptick in the mortgage book. But really it is the net of those two. So some transportation credit which arguably is muni and kind of more or less pure muni and that's offset to a degree by the -- a couple of mortgage deals.

Corey Gelormini, John Hancock.

In terms of the list of the CDOs of ABS that you talk about, can you give us some color in terms of performance of these things? And exactly how are the underlying pools doing? Obviously ,you have inputs of what the market value of these underlying things are doing, but certainly from looking at your own internal ratings, you're certainly moving much more quicker than the rating agencies on some of these things. And certainly Merrill Lynch had similar comments this morning how the market is way ahead of the rating agencies. Can you give us a sense of what performance is on a lot of these things, especially in the CDOs of mezzanine stuff, what's the underlying tranches doing?

And I guess in terms of your assumptions, Moody's just came out with what they think the ratios would be for you guys under a generic 10% subprime loss number. And I was just curious, what do you guys expect generally from '06, '07 subprime vintage in terms of losses?

Okay. Sure. So then I think the first question there was performance and what's happening. And I think in it candidly you talked also a little bit about the rating agencies. I've made the observation over the last week or so having looked in I've have to say some detail at some of the deals that they've chosen to downgrade, and it's not evident to me right now that certainly in some of the cases I've looked at, that the agencies are behind. I mean, candidly, human nature might tell you that probably they may be feeling somewhat sore and what's in it for them candidly, and I'm not trying to besmirch them in any way whatsoever, but what would be in it for them to overrate at this point in the cycle. I venture the suggestion that perhaps the pendulum is moving in the opposite direction.

In terms of performance, clearly it's difficult to generalize across the entire portfolio. And credit averages are always very dangerous. So I think what I'd point out really is the very differentiated performance across the different deals. And obviously that's most starkly seen in the deal that I think you referenced, i.e., the CDO of mezzanine ABS.

And if you look at the detailed disclosure that we give on the website, you'll see that the entrails of that give a very detailed picture of what the rating performance actually is in respect to that transaction. And you will see that the vast preponderance has been downgraded. So I'll just give you a couple of figures to put that into context. So as at the prior release of this website, so I'm turning back the clock here, 9% of the particular deal we're talking about, the CDO of mezzanine ABS was not investment grade. That number has now gone up to 60%. So you can see that there has been a very considerable downgrading of BBB collateral -- that's obviously the mezzanine. [Were it], i.e., BBB, it's the BBB tranches, the mezzanine tranches of individual deals.

As regards the rest of the high grade ABS portfolio, clearly it is a different sort of portfolio. It is a high grade ABS portfolio. And what that means is, and this is really page 8 of the website, that the preponderance of RMBS ratings is significantly higher than that for the mezzanine deal that we've just talked about. Yes, you will observe some movements in those deals. But really there has been a much smaller downward volatility, if you will, in respect of higher rated RMBS tranches. Put simply, higher rated tranches are designed to take more stress, so that is kind of what you'd expect and what you've seen.

In respect to the high grade ABS deals, we have chosen to downgrade three additional transactions, modest downgrades, but nonetheless, material ones that we obviously want to disclose. And we've obviously chosen this website portrays to downgrade three deals from AAA to AA.

But I guess the issue is from an investor from afar, just reading, not experts like you, I look at just generic delinquency rates on what's purported in the press, not obviously deal by deal, which you have, which would be helpful; you give the enhancement level. And certainly when it ultimately filters to you, it would be certainly very complex.

But I guess the issue is not knowing what the performance underlying your transactions, one can't really say whether that 22% on the first one you talk about is really good or bad. And given your previous comment on the HELOC, that you would look at it, blah, blah, high -- good FICOs, significant owner occupied, reasonable loan to values. It did lousy. And so I'm sure the issue is, is those sort of generic factors may not be good enough. And what's really going to be the tail here is actual performance.

Sure. Let me try and dissect a little bit of your comments. I think one of the points that you talked about is loss expectation. And I think you mentioned the number of 10%. I think just as a point of reference, something around a 10% cumulative loss somewhat depending on the timing of that loss is about where you will suffer the first dollar of loss on a BBB tranche. So, just as a frame of reference when you hear these losses bandied around, that's really what it means.

Clearly what we're trying to do in all these ratings is to be forward-looking. That's absolutely the intent. Clearly in relation to the CDO of the CDOs, we went through a very large exercise which we talked about on our website in terms of drilling down to individual tranches and projecting out losses that one might expect based on delinquency buckets and so on and so forth. So we are incorporating, in terms of our assumptions here, considerable stress. We're in a stressful environmental and clearly we need to incorporate that into our [ratings].

So I think that that's what we're doing. We are obviously conscious of how the world looks might now. And the fact that we are in a stressful period and our ratings are designed with that in mind. They are forward-looking ratings.

I think the other point that's interesting and I'm really focusing in here on the CDO squared products is again reference to the website, that the very different vintages that are prevalent in those transactions. And the importance of vintage is really absolutely crucial. I want to make it clear that if you can look at one thing in relation to the CDO-squared portfolio, look at vintage. And the vintage distribution is summarized I believe on page 2 of our website. That is a crucial point. So 45% of the deals are 2005 or earlier.

I should add that that isn't -- the vintage distribution is not based on a sample of CUSIPs. It's based on the CUSIPS themselves; a reference to the word sample in the paragraph above is really referring to the fixed-rate and FICO. So we're very confident about the distribution of vintage. One or two people have commented upon that; vintage is a crucial thing and is very important in terms of the sustainability of these transactions on a looking forward basis.

Mark Lane, William Blair.

I just have a question about capital. Sean, can you talk about how you're managing your capital and your capacity for new business? Obviously this quarter you didn't write any CDO of ABS or any direct subprime RMBS. So are you in a position now where you have enough concern about the broader deterioration of the mortgage market that you might pull back from writing more new business in order to conserve capital even to protect against downgrade risk? What's the thought process on managing new business growth in the fourth quarter and into 2008?

Yes. The capital management has always been dynamic process that we need to react to the environment we're in. Two major components to that level of capital holdings; the first obviously is the new business production environment and the second being what's the potential for downgrades in the portfolio, which would require a higher level of expected -- what I should call tail losses that are calculated in the rating agency models. Being that these tail losses are very extreme scenarios and they're meant to capture the vast majority of potential results here.

So we need to consider both elements in there. And there's smaller stuff like the potential for callbacks of reinsurance portfolios and the like. So for all those reasons it's prudent, and we've always carried a healthy excess capital position from the minimum AAA standard that the rating agencies would require in their capital modeling scenario.

So in this environment, clearly uncertainty of mortgages. Obviously, we've taken some downgrades in the portfolio. We're in a stressful environment. The potential for the agencies to do the same is obviously quite high. I've given you a nature of the numbers based upon if we were to go -- if the agencies or one particular agency were to go to our ratings. So that would capture a piece of the excess capital levels, not all of it, a percentage of it. And clearly we need some excess capital for new business because we think that environment is a good environment for us at this particular time.

So we've always looked at it from an excess capital position and then looked at whether or not the price has been attractive in order to make repurchase decisions and capital management decisions. So that process hasn't changed. And obviously with the two elements, new business production and the potential with downgrades, we're certainly in a conservation mode from a capital standpoint.

So does that mean that in 2008 everything, I mean everything else equal, that -- I mean the growth opportunities are there. Does that mean that you could not see your new business growing next year? Or that wouldn't be reasonable to assume at this point?

No, we've been able to, over the past several years, do new business levels and effectively maintain the levels of capital even with buying back shares. So clearly there is some ability for growth to handle those capital levels. We'll also mention that that's -- depending on the underlying business that's being written, but if you have just pricing increases in your underlying business but you're still writing the same levels of par, but you just had pricing increases, that doesn't attract more capital, if you will, from a tail credit modeling scenario; just additional premiums for the same level of capital holdings. Obviously, that's good from a return perspective but it's also good from a standpoint of utilization of capital.

So the three high grade deals that you downgraded from AAA to AA, I mean, generally, how much of the $160 million would be attached to those three transactions?

A very small amount, Mark. Most primary elements of that $160 million would be on the direct side of the portfolio where we're taking obviously case reserves for a couple of transactions and we have one transaction below investment grade which is attracting ACR. So those are going to attract bigger chunks of the money. So the downgrades from AAA to AA, while they have an impact, they have a much lesser impact than what we've done in the RMBS direct portfolio.

How much of the ACR takedown was related to Katrina-related reserves?

That amount was $27 million. And there is two elements to that. One is we had an actual paydown of the exhibition haul credit. So our overall exposures have come down fairly dramatically, almost approximately $300 million exposures to certain credits that we've been keeping an eye on. That's one element.

The other element is a broad based upgrade of credits down there in New Orleans -- not coincidentally, but the agencies, one particular agency also upgraded a number of the credits as well in the quarter.

Darin Arita, Ambac.

Actually, it's Darin with Deutsche Bank.

I was going to say, Darin --

Changed companies here. But thanks for taking the question. Turning to I guess the mezzanine CDOs of ABS and seeing how Ambac has lowered the internal ratings of these deals. Can you talk about what has surprised Ambac here? Why have these ratings moved down so quickly?

Sure. I'll take that. I think that it's more or less a statement to the obvious event. Clearly we did not expect, I think probably in common with most, we've been at the tail event that we're seeing today. Even in the papers today there's more news in respect to various folks that are active in the market. So we underwrite deals to withstand stress. That's what we do. And as we've talked about on prior calls and I think publicized in terms of some of the stresses that we run, we want our deals to be able to withstand the stress. Well, sure, as eggs are eggs, we're going through a stress now and the result is that we have had to selectively, to I think a reasonably modest degree, downgrade those deals to reflect that stress.

If we haven't underwritten to the kind of stress standards that we do, obviously I refer you to the last two columns in our website disclosure in terms of the subordination that we have got on these deals, then probably we would be, in one or two cases, below investment grade. But we're not. So I think it's an environmental thing. We underwrite to stress. We have got stress. Probably more certainly quicker than anticipated. But the good news is, and it is good news, is that the portfolio can withstand that.

I guess when we think about your ratings now, some of them are in the A category and one is in the BBB category, just how much more flexibility is there for things to get a lot worse in the housing market and for these deals to still perform?

Well, BBB obviously is the investment-grade rating. I'm not forecasting this, but if you are a non-investment grade rating, that is still some ways away from claim. Clearly as I said before, the ratings are designed to be forward-looking. We have taken these downgrades. We have inserted, if you will, some hypothetical downgrades. We have used I think appropriate assumptions in the whole modeling of these things -- correlation and other matters -- and they come out where they come out. We've checked for consistency, if you will, between different transactions and between different types of transactions to try and make sure that we do connect some of the, I'll call it more empirical work that we have done in relation to real drill down into individual RMBS securities as against the CDO modeling approach.

So we are trying to triangulate this in a number of different ways, trying to look forward in a number of different ways and use sensible and different means of analysis to come to a forward-looking conclusion. And that is what we have done. We are in a horribly stressful situation in terms of what is happening in the market. Our deals do remain investment-grade. We feel very good about that.

And to what extent do these ratings reflect the potential benefit of the various cash flow triggers in the CDOs?

That is a good question. They do. So we do model in the cash flow waterfall, if you will, in these transactions. So yes, we do look at that. And obviously, what you are doing is -- it's a transaction by transactions thing, but you model the deal including the benefit [that won't fall] obviously. In most cases, not all, but in most cases that is a positive feature to the transaction.

And just finally, in terms of -- is it the CDO-square deals -- ratings of the inner CDOs that you disclose, I believe those are really based on the lowest rating from any one of the rating agencies. I mean it sounds like, it feels like there could be a very large [swath] of rating downgrades on CDOs from some of these rating agencies very soon. And so should that happen, what should we expect further downward pressure on your ratings on your CEO-squared deals?

No, I think you raised a very good point and I am glad that you have given us the opportunity to clarify it. Obviously what we are doing in the website and I think we are pretty good at disclosing this is that these are where the rating agency ratings are right now. That isn't the same thing, it is not the same thing as the ratings that we use and the analysis that we perform in coming up with our own ratings.

Clearly, for the moment, as you correctly point out, and I think it will change very shortly now we've had all the RMBS downgrades, that there hasn't really been any significant, maybe one or two, but not significant CDO downgrades. That undoubtedly will come. What we are trying to do in our own analytics is to look through that. So very clearly when you have got a [single A] CDO-squared, you're not predicating that off the rating agency current ratings of the CDOs. You are very substantially in fact, knocking down what we believe the correct CDO ratings to be. And in fact, in respect to those transactions the CDO-squared specifically, we have increase the default time correlation to reflect some of that which we are seeing in the market.

So there is a dichotomy here between the disclosure which is rating agency ratings, and our summation ratings, that is our ratings on the individual deals which encompass our views of the respective collateral and the appropriate other assumptions.

Heather Hunt, Citigroup.

With the stock trading at about 52 and down as much as 10% today, Merrill Lynch being downgraded I think we can all agree that there are people in the market who are worried about the AAA rating. I'm not personally but can you -- you've talked about it in various ways during the course of the call -- could you maybe give us some color on the nature of the conversations you have had with rating agencies since the RMBS downgrades? Certainly it was more extensive at this point than people might have expected, which has affected the stock price.

Yes, we have, Heather, on a regular basis have conversations with all three of the agencies on a regular basis sharing data regarding the portfolio, underlying attributes in the portfolio, sharing analysis with them and obviously sharing thoughts on the overall portfolio. We obviously talk to them about any public announcements or any other things we would make there. So we have a very active dialogue with the rating agencies. And that has continued. I expect that to continue clearly over the next quarter or two over the next year to talk about RMBS type matters and CDO of ABS transactions.

Okay. And so are they -- I mean, I recognize that. I guess I just wanted to have the affirmation of that. Are they weighing in on the amount of RMBS deals that you do? Or is it just a reflection of your appetite to not do RMBS and CDO of ABS is simply that you have a pretty full plate of it now. Obviously, the pricing would be attractive. Are you choosing not to do it because you don't see the market as being disciplined yet? Or you just don't want to put investors through that? Are there any conversations with the rating agencies about that?

I haven't had conversations with the agencies directly relating to those types of issues. There obviously would be I think some sensitivities if there were to be a large number of transactions in that space with that type, say, '06, '07 collateral. It depends obviously on the nature of the transaction and protections and everything else that would go into taking those types of risks. But, no, we have not had conversations with them down that particular path.

Is it just that, A, the deals aren't really getting done and B, you don't need that much discipline in the market yet?

Well, clearly, there aren't -- I mean, as you point out, Heather, I mean volumes absolutely have dried up, both in the primary RMBS market -- I mean, if you look at some of the statistical results coming out of the big originators, year on year they're down 55% in terms of fundings. That obviously does filter through into the market. So, hey, you're absolutely right, there aren't the deals to do. But B, obviously we do have limits. We are mindful of those. We do want to see the dust settle.

That said, I think particularly on the direct RMBS side. there probably are some great opportunities to do worthwhile good risk return business and obviously we are open to do that.

In the quarter I made some comments in the script, we didn't underwrite about $2.6 billion of transactions, certain nature of the transactions we looked at as attractive. So while the primary markets have obviously been inactive as David mentioned, there are some opportunities in the secondary markets. So if the particular transactions meet our criteria considering the environment we are in, we certainly would consider that.

Great. Thank you. And just on the mezzanine CDO-squared, I don't mean to beat a dead horse to death, but can you give us some specific aspects of the deal that can give us some comfort that next month is not going to go to below investment-grade?

Heather, this is the BBB deal that we just downgraded, so this is the mezzanine of ABS CDO, correct?

Yes.

Sure. Yes. I mean I can, I mean I can tell you that somewhat repetitively I've made that 60 odd percent, [64%] or something of collateral is being downgraded. Obviously, a decent portion of the CDO collateral is a small bucket has also been downgraded. And where things haven't been downgraded clearly we have applied the downgrades that I referred to previously.

I think the other thing, and this is just good credit in my view, it is important to try and triangulate around things. So the more different things you can find of inferring information from different views and related deals and entities, the better. And certainly we have done that here in relation to some of our other transactions where we've observed collateral degradation of one sort or another.

We have obviously thought through the differences between our expectations and underwriting. Obviously we notched the deals. And what has actually happened and how -- does it appear sensible in relation to finding some kind of flaw in relation to some of the empirical work we have done in other transactions. And broadly speaking, it does. I think we have been sensibly reasonable in terms of correlation assumptions, perhaps an important element in terms of coming up with ratings on these deals. Perversely, in fact, the correlation assumptions get less important as the collateral gets downgraded because obviously the downgrading of itself generates a higher probability of loss in the first place.

But nonetheless, we have I think suitably cautious correlation numbers inbuilt into the forward-looking rating that we have. If, candidly -- and it's an if and I don't know the answer here -- if the market, the RMBS market continues to blow up to a greater than anticipated extent, then is it possible this could get downgraded? Yes, it is. But we don't see that right now and the collateral that we have underpinning this stuff is -- it's a pretty low rating. I mean this is low B type territory in relation to the average. Averages are always misleading. But that is kind of where the collateral is.

We've commented, I think, on the call in relation to the timing of losses. These things take a long time to come through. Obviously in this transaction it goes without saying, it's tripped triggers and so on and so forth. So any benefit caused by triggers -- and of course, all these transactions will very likely come in and obviously the effect here is cash flow diversion. So what that means is the structure gets delevered. That's true across all these deals to a greater or lesser degree.

So to the best of our ability we have looked at what is happening. We have looked at comparisons in relation to other work that we have done. We've hit it pretty hard. And I think a few sensible assumptions and just come up with what we've come up with. And obviously we're hoping and believing that that will stick around for a while.

Okay. One quick final question. Obviously, all of this is happening very quickly in the market. And we've had adjustments for the last few months. Is there a specific time period going forward that you are able to look out to, say six, 12 months, and sort of pull back to now? I think you said before it's about 12 months you're trying to project forward based on the current results. So that next month we don't -- I think people are just very nervous that this is the [furry] slope.

Sure. No, there is no kind of specific timeframe like that. The point I was making that as you know well, if you look at average aggregate losses across subprime 2006 right now, you will struggle to get a higher figure than 0.6, 0.7 at 1%. As I mentioned earlier on the call, above standard BBB might give you your first dollar of loss at about 10%. So this is way out there. And obviously what you've got to recognize here in terms of timing -- and clearly that is an important point, vis-a-vis our business model in the absence of liquidity risk, et cetera -- is that this is looking forward. There's not a great prospect I think of paying claims on the underlying RMBS deals because of the slow accretion of losses over a number of years.

What ratings are doing, and I do feel and as I mentioned earlier, that just perhaps the pendulum is swinging there in terms of the prudency or aggression of those ratings. I think it is swinging a little bit in terms of what I've seen, what ratings are doing is looking at forerunning data; so, looking at delinquencies. I can see and observe downgrades in agency ratings with more or less 0 cumulative loss on fully funded OC and some sub bonds. Now that is a position you can take. Clearly you have to feel pretty strongly about that to downgrade to a significant extent. That is beginning to happen. And what you are seeing here is in some part the result of that.

So the question you need to think about and obviously we're thinking about also is the difference, if there is one, between reality, what's really happening in terms of credit, and what's happening in terms of ratings, which are forward views of credit. And obviously, what you're doing is putting those ratings in the plot to a reasonable degree with some amendments and coming out with the ratings that we are exhibiting. We feel that we are being sensible and trying to look forward in relation to the ratings that we're putting out into the market.

Tamara Kravec, Banc of America Securities.

Just a couple of clarifications. The downgrades that Moody's did and the downgrades that S&P did, what proportion of those did you say was captured in this quarterly information?

What we did was to encapsulate -- so we had to have a cutoff date. That was basically the problem. So in relation to what you see on the updated website, the cutoff date was last Friday. So that caught Moody's, didn't catch S&P; obviously, as I said before, highly likely to be some overlap there.

Just to reiterates though in terms of the analytics -- so I'm differentiating what's on the website from the analytics that have gone into the Ambac internal ratings -- we did incorporate some adjustments, as I mentioned before, in relation to those deals, 2007 deals, which haven't yet been hit, for want of a better word, by rating downgrades.

Okay. And you had mentioned I think that a few deals have tripped triggers and then will be expected to trip triggers. But as a percentage -- and maybe this is hard to answer -- but as a percentage of all of your CDOs that you've written, do you have a sense of how many are starting to trip some triggers?

Yes, I do, and I'm going to slightly look forward here. I mean I think of the 28, I believe it's six currently, that's a bit misleading because obviously we just had all these downgrades. And therefore I think our best guess -- and there is a margin of error on this clearly but just to try and give you a sense here -- I think that that six number will rise to low 20s. So I think we can expect in very short order, the vast preponderance of these things to hit their triggers.

Okay. And are there are multiple triggers on each deal? Or does it depend on the deal?

It depends. But broadly speaking, the batting order so to speak is the deals go sequential, they stop reinvestment, and at some point you obviously get the opportunity, if you should want to avail yourself of it, to replace the manager. So it is a kind of progressive thing. And these triggers I'll hit, just so we're all clear, by the application of OC ratios, which in effect hair cut the downgraded collateral, which means that the ratio obviously deteriorates, goes through various levels. And there are kind of ascending orders of control or things that the senior class, i.e., us, can do.

Okay. And what is the best way in your view to track the deterioration in your credit? Is it looking at the ACR, is it progressing up? Or is it getting a sense of whether we are at the bottom of the downgrade or -- what in your view -- because it's hard for us to see the triggers going off. Is it that we should have some confidence in your visibility into the collateral in the CDO-squared particularly, and then we're just kind of watching your reserve activity as you're going through the motions?

I think that the best single thing that you can do obviously is look at the ratings that we're putting out there. We will say that obviously I think you make a good general point in terms of observation. One of the things that we've noticed candidly is that trustees are under great pressure. They've obviously been inundated with the explosion of structured finance products. And some of it obviously isn't performing particularly well. So one of the things that we're having to do just on the theme of observation is to make sure that we know what the trustees should be doing. And I can tell you that in two cases in the last two or three weeks, we had severe questions about that. And one very good spot by one of my colleagues will in effect mean that one of our exposures will be levered by about I think it was $29 million.

So everybody's got to look at everything. We certainly are. We have a lot of people obviously looking at this. I think that storm will pass as far as we're concerned to some degree because these things will, given the [failings] of downgrades that we've had, go to a much more steady state, sequential amortizing type basis. But in the meantime, we're watching it like hawks and I think that the best thing that you can do is, in summary, obviously comb the website and obviously get to what our ratings are; they encapsulate our views of our credits.

And then just add to David's comments is clearly we've been talking primarily about the CDO of ABS. We also did, as you know, the detailed disclosures relating to the RMBS direct portfolio. Also that there is current ratings there as well by transaction.

If you really want to get into it, obviously on our website I think is a wondrous beast. And you can sort by vintage, by product, by a number of things. And if you're handy with an [entex] machine, you can have a -- you can while away many happy hours. And I'll leave that to you.

All right. And then another question on the CDO mezzanine deals that you've done. You know there's been I think some speculation and chatter about some of those being worthless, is the term that I often get. And what do you say to that?

I'm sorry, I didn't hear the term.

Yes. Worthless. That, you know, you get phone calls and there's a panic level out there that the CDOs are termed to be worthless. And I think the differentiation falls in worthless in the market, right. The market value that's assigned to these is very low, but the collateral is performing fine.

Well again, our view is simply encapsulated in the investment grade rating. We are not in the habit of putting investment grade ratings on worthless securities. That's not where we're at.

Right. So your ratings are really the best indication and as those progress downward into the low investment grade, your ACR goes up and your reserves are moving up.

Well, I hope that we won't see that progression, clearly. But if that was the case, that might in fact [be going up].

I'll just correct you a little bit there. On the ACR and loss reserving, those are for insurance contracts.

Right.

I did (multiple speakers) --

Not the CDO, right?

Yes. When we're dealing with credit derivatives what we would do is we would adjust our operating earnings accordingly. So therefore if we thought that it was necessary to reflect reserves, if you will, we would be adjusting operating earnings. But from a financial reporting perspective since they're derivatives, we're estimating the mark-to-market.

Mike Grasher, Piper Jaffray.

Questions were asked and answered. Thank you.

Andrew Wessel, JPMorgan.

Yes, I had a question on I guess a couple of things. What kind of look-through do you actually have to the underlying collateral in CDO-squared? I know that in a lot of the underwriting for those initially is manager level, especially on deals that have ramp features and reinvestment features, obviously you can't underwrite 100% of the collateral anyway. So just give us an idea of what kind of look-through you have there.

Sure. Well, candidly, a very extensive one. And we summarize what we've done in relation to one of the two ways that we look at those deals on the website. So, from memory, I think there were 15,316 CUSIPS. So we've drilled all the way down there. We've looked at roll rate analysis CUSIP by CUSIP, obviously taking into account the particular deal. We have played around with loss curves in respect of all those CUSIPs to look at the potential for reset risk or the acceleration of front [head] losses. And we have done all of that using various different forms of roll rates. We painstakingly insert the correct RMBS deal into the correct [in a ] CDO. We then painstakingly put the [right in to see the right out] of CDOs.

So there is a great amount of look-through that you can do. It is a vast computational exercise that we undertook a month or two -- a month or a half ago, whenever it was. Obviously, we'll be repeating that. And hopefully enhancing it in some way or other as we go. So we're not overly reliant upon, if you will, standard CDO models of rating and the like. We have done drilled down.

I will say however, just the completeness that we obviously do look at these deals in addition to on a drill down basis on a more standardized CDO method, which I think is probably what you're referring -- obviously that is the whole triangulation thing and let's look at any particular problem from as many perspectives as we can. Clearly what we are looking for is information and doing that, are they consistent? Are they inconsistent? And so on and so forth.

So in answer to your question, a great deal of disclosure. We know the CUSIPs. We know where they are. It's absolutely empirical. It's vintage dependent. It's distribution dependent within the inners and the inners to the outers and we can do it that way and we have done it. And we will continue to do it. As a secondary check or a triangulation, we use more standardized models to see if we can glean any useful insights in doing so.

Thanks. In general, what kind of HPA assumptions do you put into your analysis? I mean obviously, everything is geography based and subprime properties have been shown to decrease more due to foreclosures and vacancies than prime properties. I understand all that. But just in general, what kind of HPA assumptions do you use?

I'm going to assume -- but correct me if this isn't the case -- that you're referring to the CDO-squared product?

Well, no, I mean, just in general when you're underwriting any mortgage bond. I mean if you've got to work through to the CUSIPs --

Let me take the two in turn. I will start off with the CDO-squared analysis because that's useful. And I'll move on to our direct mortgage book.

So to kick off then with the CDO-squared analysis, what we have done to look at CUSIP by CUSIP, the individual collateral performance, is a roll rate analysis. Essentially, that is trying to understand the propensity for borrowers within a particular RMBS deal to roll from one bucket, i.e., delinquency bucket, 30 days to 60 to 90 to foreclosure to REO, et cetera. So you can look at methods of viewing that propensity, estimating that propensity. And obviously at the end of the line what you get is foreclosures and REOs, and obviously shortly thereafter loss. And the question I think you're asking in one sense -- and the way I am going to choose to respond to it -- is what is the severity of the loss once you've got to the end of the line. And what we've have done -- and I think you'll feel it's pretty reasonable -- is assume a severity of 50% in relation to the analysis that I've just described. So loss severity on the mortgage loans of 50%.

That compares -- obviously, if you look at the data and look at data over a number of years, you will find that the data supports the notion that early on in vintages severities might be 20% or 30% and then they'll gradually rise through time to maybe 60%, 70% or somewhat higher. We're assuming 50% flat throughout. So that is what we have done on the CDO-squared book.

If I turn now to the direct book and how we underwrite the direct RMBS deals, that describes you what we do. We view loss as a combination of the frequency of defaults and obviously, the severity of default. We look at the frequency of default as defined by things like the class of the borrower, the LTV, and so on and so forth. When we look at the severity of the default on a completely normal transaction, what we would assume is that house prices fall 30%. So we're not going MSA by MSA; we're saying right; everything perfectly correlates it, falls 30% that happens to default. And then we work out the loss on that basis.

If we have a particularly concentrated pool, California being an example, then clearly what we want to take account of is the potential for regional concentrations if you will, I guess that is behind some of your question -- what we'll do in those circumstances is to up that severity number by increasing the house price decline. So for example, we might choose a figure of 40%. So we are trying to capture the severity issue in a number of ways. And feel by and large that we do.

Okay. Great. Thank you. And then my last question is just to clarify -- so based on the fact that nothing is internally rated in your CDO portfolio is below investment-grade, at this time there's no reserve for that and that really the economics have just been captured -- not the economics -- but it's been captured accounting-wise -- just the mark-to-market?

Correct.

Terry Shu, JPMorgan.

There have been quite a number of questions. Let me just -- again, I don't want to beat it to death, the $510 million deal that was downgraded several notches. David, if you can explain again -- I had thought that when you underwrite a deal you always use your own internal ratings. But in this case the downgrade was in part triggered by the rating agency's downgrades. Can you explain that again? Do you not do drill down analysis from the point of underwriting?

Sure. What we generally do, what we always do is at underwriting we look at obviously the component securities. And we assess whether or not we think that those securities are appropriately rated. And in this and some other cases we decided that that wasn't the case. And we notched those securities, i.e., we chose to input for the purposes of the internal rating that we ascribe a rating lower than the then-existing agency rating. That's what we did at underwriting.

But now I turn the clock forward to where we are now. And what we have observed is that there has been a pretty hefty element of catching up, if you will. And in some cases regrettably surpassing our expectations of downgrade that the expectations that we had at underwriting. So in some cases, yes, the deal that you referenced in particular, what's actually happened is that despite the fact that we notched very considerably, we in retrospect according to current agency views and the prospective view that I've summarize previously, didn't notched enough. And on that basis obviously we've applied those collateral ratings and so on and so forth and come up with the ratings that we have.

So it was not done on a CUSIP level at underwriting then?

No, that's not true. It's done on a deal by deal basis. Taking into account what we felt about the issuer. For example, if we tiered originators. So if we happened to think that one particular originator is really going to struggle and could go out of business or something, then one of the things we might want to do is effectively write that issue, that collateral, off. Economically we do that by obviously getting more subordination as you can see on the website.

So we did do it deal by deal. I underline collateral by collateral at underwriting. But in this particular case what has happened is that the agency downgrades plus this perspective methodology that I outlined have surpassed even our somewhat conservative expectations at underwriting.

Right. Okay. And now a lot of the analysis you're doing, as you said, is very painstaking because it takes so much time, is even more granular just to give yourself extra assurance.

That's right. We progressively have to up the ante and we look at this often and obviously we are doing. It is interesting to note that even in a few cases and I think I referenced number four, I think it is in fact in five cases, despite the tremendous stretch that we are seeing, and we can't understate the fact that this is a horrendously stressful environmental, nonetheless in five cases with respect to our high grade deals, the ratings now are in fact higher than that which we considered that they might be at underwriting.

Now clearly at underwriting we did not consider -- or considered that we did not expect the horrendous environment that we've got, but we were thankfully [superly] pessimistic and therefore can still feel good about investment-grade ratings.

One more clarification. As you said, a lot of the losses will not be paid for an extended period of time. Those particular HELOC deals, the two RMBS deals where you put up case reserves, I believe it was mentioned that you have paid a claim on one of them. Why did the paid losses come up so quickly?

That is the particular structure of the deal. So what I tried to outline a few minutes ago is the fact that unusually and obviously it's not happening now, the structure of that deal didn't have OC to begin with. It was a deal, it was called even though it's a build up deal. The particular deal stated at zero, it's supposed to rise to I think it was 4%, 4.1% of OC. But because of an OC holiday -- an OC zero balance to begin with then a holiday, and what happened is losses came in so quickly and the performance was so poor, they obliterated the gross excess spread, which meant that OC didn't have time to build up, which means you're then immediately paying a loss.

Okay. It's just that it's so soon versus years away. It seems like it's quite a gigantic difference. I'm right there, right?

Absolutely. It is an unhappy idiosyncrasy of poor performance and structure that wasn't appropriate for the deal.

Third clarification is when you talked about the CDOs severity test that you used 50%. And then you talked about your direct RMBS using the 30% immediate home price decline. That is not apples-to-apples, right? Because home price decline, then you have foreclosure, there is also additional cost in terms of time and how the banks have to pay insurance costs or other expenses. So are they apples-to-apples? The 30 and the 50? Or are they not apples-to-apples? Because you'd think you would have home price depreciation and then on top of that, other costs.

You are absolutely right. Obviously 50% was the overall severity, 30% is the house price decline. I will say, and I skipped over this detail, thank you for correcting me on it -- obviously we do factor in ancillary costs when we do a foreclosure and so on and so forth -- when we do the direct business. Let me outline for you what those costs are.

It is things like the carry costs. Clearly you are carrying the mortgage, you're not getting any cash from it. Typically if somebody has left the house, they may not pay real estate taxes. As senior, you will have to. Typically there will be costs of foreclosure. The costs of foreclosure are kind of interesting and quite detailed. So for example, if you have a property in California, you can actually get through the whole process very quickly. That means less carry costs, probably less legal costs.

If you are in New York, actually reverse the case -- it takes a great deal of time. So what you have to do and we do do this, is to factor in what loans are where and so on and so forth. If you typically add up these ancillary costs, and I'm not going to say it doesn't vary across location and time and all the rest of it, you will get to something like 20%.

So the 30% is consistent with the 50% -- broadly?

Broadly. It is not apples-to-apples but it really isn't a million miles apart.

Where does loan to value come in, in that analysis? Is than an adder or is that included in the 30%?

Loan to value obviously is included. Because -- in two ways. A, it will be a determinant of the frequency of foreclosure. Obviously, the common sense being that if you have no equity in your own house you are more likely to default, so it will up the frequency.

In terms of the impact upon severity, clearly if you are 100% LTV, then a 40% decline in your house is going to knock 40% of the mortgage. If you are at 90% LTV, then it won't. In other words, there is some equity. And obviously we therefore in the computation which by the way is loan by loan on the direct business, will build in the actual LTV of the loan to the severity of the particular loan (multiple speakers) --

Right. But it is a correct description to say that the 30% is a home price depreciation assumption.

Yes. We occasionally use a higher number for that and that is a concentrated [force].

Thank you, because people talk about home price depreciation of 10%, 20% over the next two years and how that could trigger much worse performance. But I gather the other big element of course is recession concern, which impacts the delinquency rate.

Yes, I mean just to be absolutely clear, so we are 100% clear, I don't want to misrepresent or mislead. What I am talking about is the application of a house price assumption to severity. What I haven't talked about is the application of house price falls, forward-looking LTVs which might then become a frequency matter. We take the frequency at outset. The severity is determined obviously by the house price and all those other things.

Right. On that front, if you can comment, because I think one of the concerns is in the event we are in a recessionary environment, what happens, and also the behavioral response is depreciating home prices. I assume that none of that has escaped you, that you have tested for that and you are comfortable that it can withstand some reasonable scenario that we could have some economic weakness that you'd didn't build in, that everything will be just fine. I mean there could -- correct?

Yes, that's right. These things are investment-grade. They're designed to withstand stress. Clearly we have an expectation of loss models loss. And we need to make sure that the announcement in the deal is some multiple of model loss.

Right. Can you then one additional clarification then -- the frequency assumptions, can you talk about that broadly? Or is that too difficult to talk about broadly?

No, broadly it is easy. And detailed it is very difficult. But the notion here is that the frequency of foreclosure is determined really by a couple of things. One is the I've broadly described as the nature of the borrower. Is he high FICO? Is he a grade A type person or a great C type person? And that will, other things being equal, be an input to the frequency.

The other main component is CLTV -- loan to value. So clearly, as we said before, if you have a 100% LTV, you might expect him not to try quite so hard to stay in his house as a 40% LTV.

Right. How about economic environment? Is there some way to exclusively build that in?

There are various ways that you can do that. But I think the problem is with some of those models they're not great at turning points. They rely on statistical histories and histories are great when the market is going up and so on and so forth. We do periodically use one such model. The difficulty is, candidly, in calibrating that model successfully. Really what you're affecting on those is to return to the kind of base analysis, if you will, is that you're calculating the model loss and doing a transaction at the multiple of that model loss. That if you will is the insurance policy that means that the individual transactions are investment-grade.

Right. And you always do that with multiples?

Absolutely. We do not want to do a deal (multiple speakers) --

Two times or something -- it's hard to be -- but two times would be a good round number.

Yes, for a BBB type deal, yes.

Rob Ryan, Merrill Lynch.

Could you comment on the new business environment in the context of first of all, the overall spread widening on securities? What portion of that spread the bond insurers might be able to capture based on competition among the companies themselves, as well as how insured paper is holding up?

There seems to be this contradiction between some absolute doom and gloom expectations out there for untold levels of huge losses for bond insurers. And then again the general fixed income community's strong demand for the product.

Sure. Let's start with the nature of the market and kind of give some color there. Clearly on the pricing side where we're seeing that is clearly on the structured finance sector of our book. And is really two elements of the story there that we've seen this quarter. We've seen it clearly on the pricing side. And we're seeing it a bit on the penetration side. And that is the widening of spreads allow us to participate at risk reward dynamics that we find attractive.

And one good example of that, which we mentioned, was the student loan transaction that was closed during the quarter. That was to an issuer that has typically, at least over the last several years, utilized the senior subordinated market and it was a transaction that we did not participate in. So that in and of itself was a good thing. And we are seeing stuff like that.

On the pricing side, we did write some mortgage transactions; pricing levels there are much higher. We were still able to close some CDO transactions on the corporate side and seeing much better pricing in that area. Multiples of what we've seen over the last several quarters. So really two pieces of news there.

When you look more broadly to public finance and international, you're not seeing the same price contagion. But on the public finance side, you're still seeing strong issuance. It looks like the year is going to set a record pace for issuance overall. Still good penetration levels. And we saw, at least in this quarter, we saw some uptick in the types of transactions that we like to write; saw some of it in the housing sector and the health care. So, on the military housing side, that is.

Also closed some big deals too, which had some very attractive returns on them. But what we're seeing there, Rob, on the pricing on the public finance -- still stable, characterized by competition and available spreads being effectively stable. So really not a pricing story; more of a volume and a mix story there.

On the international side, we're starting to see some modest price contagion internationally. But really didn't see much in the third quarter. So we're hopeful some of that carries over. But still lumpy internationally. We obviously closed some big deals; had a major impact on the portfolio. So we're still hopeful that some of the price contagion will cross the pond, if you will, and we will see some of that come through in the production numbers.

Great. Thank you very much.

On the insured paper -- I have the second part of the question -- on the insured paper, certainly we haven't seen, at least amongst the major model lines, we haven't seen -- you see trading levels fairly consistent, even though the credit default swap market prices might be a bit diverse across the guarantors. But haven't seen that impact. I think one would have to look at that vis-a-vis the change in available spreads, particularly in the structured finance markets. So we were able to see some favorable overall pricing with those two dynamics effectively in concert with each other. And we haven't seen that dynamic on the public finance side having an impact.

Adam Starr, Starr Asset Management.

I didn't think you'd get to me. Quickly, if you were to do the mark-to-market on the swaps today, would the spreads be meaningfully different from where they were at the end of the quarter? Or is it hard to tell?

And second of all, when your mark is on what the swap would cost to do, if I understand correctly, [done], what's the underlying asset would (inaudible) for, if you could explain that?

Yes. Both are correct. We haven't repriced the portfolio. That's a process that we undertake. I will make an observation though, it seems the indices, which aren't necessarily the perfect predictor but at least directionally have been decent predictor, have remained relatively stable, as best I saw late last week. So no predictions. We do think that there's potential for volatility in that number. So we'll just have to keep a close eye on that.

And then the second question, could you just refresh my memory on it?

Would some of the difference between what Merrill did and the degree of mark that they talk and the degree of mark that you show be that you are mark -- and I may be understating this. You are marking the cost of doing a swap today as opposed to the market price of the underlying assets?

Right. Right. Yes, what we're -- the contract that we have is the contract as stated in the typical case with an investment bank, over the counter contract, with certain selected terms heavily modified, tailored, has specific attributes to it. So under the accounting rules we're looking to mark that contract. But a key component of trying to figure out the value, or at least the estimated fair value of that contract is to understand what's happening with the underlying reference pool or reference bond, whatever the case may be.

And the way we select to that is to obtain underlying prices from the arranger bank that was involved with the original transaction that's obviously knowledgeable about that particular transaction. So that's a third party source that goes into the process of estimating the premium levels that that contract would gather.

But you're taking a mark on a different entity. It may be a derivative of the first entity, but it is not the same thing.

It's not the cash bond. The actual cash -- if there is a cash bond obviously influences the level. And the level, but it is one component to the process.

We've come to the end of our time. I would like to turn the call back over to management for closing comments.

Well, thank you very much for all the active questions. We do acknowledge that there were additional questions in the queue. We apologize for not getting to all. We did carry the call over a bit longer than we would typically have. So we are available. Pete and myself clearly are available to answer all questions so please call us directly. And again, thanks for attending our conference call.

This concludes today's teleconference. Thank you for your participation.