July 1, 2010
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
United States Securities and Exchange Commission
Division of Corporation Finance
100 F Street, N.E.
Washington, D.C. 20549
Re: | Ambac Financial Group, Inc. |
Form 10-K for the fiscal year ended December 31, 2008
Filed on March 16, 2009
Form 10-Q for the quarter ended June 30, 2009
Filed on August 10, 2009
File Number: 001-10777
Dear Mr. Rosenberg:
This letter is submitted on behalf of Ambac Financial Group, Inc. (“Ambac”) in response to comments received from the Securities and Exchange Commission staff (“Staff”), in a letter dated April 29, 2010 (the “Letter”), with respect to Ambac’s Form 10-K for the fiscal year ended December 31, 2008 filed on March 16, 2009 (“2008 Form 10-K”) and Ambac’s Form 10-Q for the quarter ended June 30, 2009 filed on August 10, 2009(“June 30, 2009 Form 10-Q”). The Letter is a follow up to our response letter dated March 12, 2010, with respect to Ambac’s 2008 Form 10-K and June 30, 2009 Form 10-Q.
Ambac appreciates the efforts of the Staff in this review process. Enhancement of the overall disclosures in our filings is an objective that we share with the Staff and one that we continuously consider in our filings. In connection with responding to your comments, we acknowledge that Ambac is responsible for the adequacy and accuracy of the disclosures in our filings; that SEC staff comments or changes to disclosure in response to SEC staff comments do not foreclose the SEC from taking any action with respect to the filing; and that Ambac may not assert SEC staff comments as a defense in any proceeding initiated by the SEC or any person under the federal securities laws of the United States.
In each of our responses below where we indicate we will provide revised disclosures in the future, we will begin including those disclosures in the June 30, 2010 Form 10-Q.
Securities and Exchange Commission Staff Comments:
Form 10-Q for the Fiscal Quarter ended June 30, 2009
Item 2. Managements’ Discussion and Analysis of Financial Condition and Results of Operations
Critical Accounting Policies and Estimates
Borrower Default Burnout, page 65
1. | We understand from your response to comment 1 pertaining to CDS that it has not been feasible to pursue the remedies described in your response and therefore you have not recognized contingent gains that may stem from these remedies. Please confirm our understanding. If you ever reasonably expect to record contingent gains that may stem from these remedies either as an asset prior to its realization, or as a material component of CDS fair value, we request that you discuss the matter with us before filing. |
July 1, 2010
Page 2
Ambac response:
The Staff’s understanding is correct that we have not recognized contingent gains stemming from remedies described in our previous response letter dated March 12, 2010 related to the third bullet point of comment 1 in that letter. We will contact the Staff, as requested above, if we ever reasonably expect to record contingent gains that may stem from these remedies either as an asset prior to its realization, or as a material component of the CDS fair value.
2. | It remains unclear why you are not including interest and/or fees for loans which have not been liquidated or charged off in your subrogation recovery calculation, using the random sample or adverse sample approach, for those securitizations where such amounts are part of the sponsor’s contractual obligation. Please tell us why you are not estimating these amounts in your calculation. |
Ambac response:
Based on follow-up discussions we had with the Staff the week of May 17, 2010 and on June 10, 2010 it is our understanding the Staff has several specific questions related to comment 2 above, with respect to Ambac’s overall loss reserve estimation process as follows:
• | Why are we able to estimate interest as a component of the gross claim liability (i.e. cash outflows) but unable to estimate interest as a component of the subrogation recovery (i.e. cash inflows)? As a follow on to this question, what is our interpretation of the applicable accounting literature for subrogation recoveries where we do not believe we have adequate information to develop a reasonable estimate? |
• | Are we estimating the gross claim liability outflows and subrogation recovery inflows independently of one another for purposes of discounting these cash flows? |
• | With respect to the random sample approach only, does our methodology overstate the subrogation recovery by including loans that we believe have breached representations and warranties but which we do not anticipate actually defaulting? As a follow-on question, how does our subrogation recovery estimate differ from the loan repurchase mechanics as described in our current disclosures? |
Regarding the discussion below, we refer you to the following RMBS loss reserve disclosures in our March 31, 2010 Form 10-Q: i) pages 72 through 78 for a discussion of the gross claim liability estimation process and ii) pages 78 through 82 for a discussion of the subrogation recovery estimation process.
With respect to the first question of the first bullet point above, the analytical models we use to estimate the gross claim liability (i.e. cash outflows) are based on either i) for the second-lien transactions, a roll-rate methodology that observes and projects relevant trends in the mortgage pools (i.e. delinquency rates, default rates, loss severities, etc.) or ii) for the first-lien transactions, a cash flow model that projects individual cash flow scenarios using a Monte Carlo methodology at the loan level. In both methodologies, relevant data such as the weighted average coupon of the mortgage pool, weighted average life of the transaction and cash flow waterfall of the transaction is readily available to enable us to project both principal and interest shortfalls in the transaction, which are then discounted back to compute our gross claim liability. However, with respect to the subrogation recovery (i.e. cash inflows) estimation process, to the extent a sponsor is required to pay accrued interest as a result of repurchasing a loan, that interest only accrues through the charge-off date or liquidation date of the individual loan. Therefore, in order to estimate the accrued interest component of the subrogation recoverable we would have to estimate
July 1, 2010
Page 3
the date at which each individual subrogated loan is charged off or liquidated, which we do not believe is reasonably estimable since the timing of such an event is at the discretion and control of the servicer.
With respect to the follow-on question of the first bullet point above, ASC Topic 944 addresses the appropriate accounting for financial guarantee contracts and incorporates the guidance in FAS 163 (pre-codification). Under this accounting guidance a claim liability is measured as the present value of expected net cash outflows to be paid under the insurance contract discounted using a current risk-free rate. Expected net cash outflows (cash outflows, net of potential recoveries) are probability-weighted cash flows that represent the likelihood of all possible outcomes using the insurance enterprise’s own assumptions based on all the information available to the enterprise. As previously discussed, we do not believe we have the information available to estimate the interest and fee component of the subrogation recoverable. Furthermore, the background section of FAS 163 suggests that the concept of evaluating all possible outcomes is subject to practical considerations. Specifically, paragraph B34 of FAS 163 suggests that information an enterprise uses for risk management and surveillance purposes should form the basis for the claim liability recognition and measurement approach. Additionally, paragraph B38 of FAS 163 states that the claim liability measurement approach is subject to a cost-benefit constraint which is similar to the one described in paragraph 51 of Concepts Statement 7. Paragraph 51 states “Like any accounting measurement, the application of an expected cash flow approach is subject to a cost-benefit constraint. In some cases, an entity may have access to considerable data and may be able to develop many cash flow scenarios. In other cases, an entity may not be able to develop more than general statements about the variability of cash flows without incurring considerable cost. The accounting problem is to balance the cost of obtaining additional information against the additional reliability that information will bring to the measurement. The Board recognizes that judgments about relative costs and benefits vary from one situation to the next and involve financial statement preparers, their auditors, and the needs of financial statement users.” We believe we have been using the best information we have available to estimate the subrogation recoverable and that this information is consistent with information we use for risk management purposes. Furthermore, we do not believe we can determine a reliable amount related to the interest and fees component of the subrogation recoverable without incurring additional costs that would likely outweigh the additional level of precision we might achieve in refining such an estimate. Consequently, we believe our current disclosures regarding the components of the sponsor’s contractual repurchase obligation included in our subrogation methodology are appropriate.
With respect to the second bullet point above, we are in fact estimating the gross claim liability outflows and subrogation recovery inflows independently of one another for purposes of discounting these cash flows. Using a simplified example, if we estimate our gross claim liability outflows will be $100 per year for 10 years (equal to $1,000) and our subrogation recoverable inflow will be a lump sum of $600 received in 3 years, we calculate the present value of those cash outflows and inflow as separate payment streams. That is, we do not reduce our cash outflows from year 3 and going forward for the inflow received in year 3 and then discount the net cash outflows. The reason being there is uncertainly with respect to the settlement mechanics of any subrogation recovery but whatever the outcome, our obligation under the policy is not extinguished and our policy is irrevocable and must remain in effect. For example, the settlement could be remitted directly to Ambac, in which case we would still be obligated to make ongoing claim payments, and therefore it would be appropriate to discount the cash outflows and inflows independently. Another possibility could be that the settlement proceeds are placed into the securitization trust of the specific transaction to pay future claims, in which case our future claim payments would be effected by the settlement inflow for purposes of discounting the net claim liability. Nonetheless, due to this uncertainty we chose to discount the cash outflows and inflows independently. Please refer to the attached Exhibit for revised pro-forma disclosures to clarify the possible mechanics related to a subrogation recovery settlement.
July 1, 2010
Page 4
With respect to the third bullet point above, as discussed in our current disclosures and previous response letter to the SEC, the “breach rate” in the random sample approach is applied to the sum of i) the CULPB and ii) realized losses resulting from loan liquidations or charge-offs, to estimate the subrogation recoverable. The CULPB population does include loans which may not have been projected to default in our gross claim liability estimation process. Nonetheless, we do not believe we are overstating the subrogation recoverable estimate. As we have previously disclosed in our SEC filings, in a typical discrete loan by loan repurchase, the repurchase price is paid by the sponsor to the securitization trust which holds the loan. The cash becomes an asset of the trust, replacing the loan, and that cash is applied to the trust indenture payment waterfall. Therefore, we acknowledge that a sponsor’s repurchase of a performing loan would theoretically have little or no impact on Ambac’s future claim payments. However, this only describes the repurchase mechanics in discrete situations where the sponsor willingly acknowledges the breaches of representations and warranties and agrees to repurchase those breaching loans. With respect to the transactions for which we have recorded a subrogation recovery, we believe the breaches of representations and warranties are pervasive and these transactions are in fact, subject to an overall transaction-level remedy which will result in subrogation recoveries which we have limited to our ever-to-date paid losses plus the present value of projected future paid losses (combined, referred to as total incurred losses). Specifically, we believe it is unrealistic and impractical that a judicial court would require the sponsor to re-underwrite every individual loan in a transaction to determine which ones have breached representations and warranties and require the sponsor to individually repurchase those loans. There is established precedent for the utilization of a statistical sampling and extrapolation methodology across a population to prove liability and damages where it would be impractical to make a determination on an individual basis. Rather, we believe the court would likely require a monetary settlement based on a reasonable methodology, such as our random sample approach, and limit such a settlement to Ambac’s incurred losses. That is, the settlement monies would be either remitted directly to Ambac, placed in the securitization trust, or otherwise held under an arrangement for the benefit of the securitization trust, however individual loans would not be repurchased from the trust. In either case, those settlement monies would reimburse Ambac for past claims and also offset future claim payments. Consequently, since the sponsor is contractually obligated to repurchase those loans which breach representations and warranties regardless of whether they are current or defaulted, we believe the appropriate measure in estimating subrogation recoveries is to apply the breach rate to both performing and defaulted loans. Please refer to the attached Exhibit for revised pro-forma disclosures which better distinguish between the sponsor’s contractual repurchase obligation on a loan-by-loan basis vs. our estimate of subrogation recoveries based on an expected transaction-level settlement with the sponsor. Per our previous phone conversations with the Staff we have also included in the pro-forma disclosures some enhanced discussion regarding how tranches insured by Ambac are impacted by subrogation recoveries from the sponsor.
3. | Please refer to paragraph (iii) on page 7 of your response and address the following: |
• | Please tell us why you are not estimating a subrogation recoverable for the other 77% of the impaired loans in the population in light of your belief that there is an ample population of additional impaired loans where breaches of representations and warranties exist. |
• | Your disclosure appears to justify that no realization factor was applied to the adverse sample because you believe additional loans exist that may be successfully challenged in negotiations or litigation with the sponsors. Your disclosure in the random sample approach incorporates a realization factor, which is your view of the uncertainties surrounding the settlement negotiations and litigation processes. It is unclear as to your basis for not applying a realization factor to the adverse sample. Please advice. |
July 1, 2010
Page 5
Ambac response:
We would like to clarify that in our March 12, 2010 written response to the Staff’s comments on this topic we indicated that we had re-underwritten 23% of the entire impaired loan population at September 30, 2009, which also represented 3% of the total loans (impaired plus unimpaired) within the pool. These percentages were calculated based on the loan count, i.e. thenumber of loans in the pool. When we first modified our actual disclosures in the December 31, 2009 Form 10-K, we disclosed that we had re-underwritten 40% of the impaired population which also represented 4% of the total loans in the pool. The reason for this difference is that the December 31, 2009 percentages were calculated based on theoutstanding principal balance of the loans rather than the loan count. We believe using the outstanding principal balance is the more appropriate way to calculate and disclose these percentages. In our response below we will be referring to these percentages based on the outstanding principal balance, consistent with our actual disclosures.
Based on follow-up discussions we had with the Staff the week of May 17, 2010 and on June 10, 2010 it is our understanding that the Staff wants to understand if there is contradiction in our reasoning when we consider how to treat the loans not re-underwritten in the impaired population and whether to apply a realization factor to the loans which have been re-underwritten. We understand the Staff’s more detailed questions are: i) have we effectively estimated subrogation recoveries on the 60% of the impaired pool that were unexamined by not applying a realization factor to the 40% that was re-underwritten, ii) if the non-application of the realization factor is not a statistically valid approach, how is this consistent with our original reasoning on why we did not attempt to extrapolate to the other 60% of the impaired loan population and iii) have we misstated our financial statements based on how we applied the adverse sample approach?
To briefly summarize, our reasoning for not estimating a subrogation recovery for the other 60% of impaired loans which were unexamined is because the impaired loans that have been re-underwritten do not constitute an appropriately selected sample to enable a statistically valid extrapolation to the entire impaired loan population. Our reasoning for not applying the realization factor to the 40% of impaired loans thatwere re-underwritten is because it represented a small proportion of the total impaired loans and an extremely small proportion of the total loans (4% of impaired plus unimpaired) in the pool. Thus we believe that for any breached loan we identified in the re-underwritten population that might be successfully challenged by the sponsor there would be other breached loans that exist in the remaining 96% of the unexamined loans in the total loan pool.
With respect to question (i) above, we acknowledge there is validity to the Staff’s viewpoint that by not applying a realization factor to the 40% of the impaired loans that were re-underwritten, that we have effectively estimated subrogation recoveries on the remaining 60% of the unexamined impaired loans. That is, we acknowledge there are uncertainties surrounding the settlement negotiations and litigation processes and that would normally warrant a realization factor being applied, but we did not do so because we believe subrogation recoveries will arise outside the 40% of the impaired population that we re-underwrote because of the pervasiveness of the breaches found within our sample. More specifically, we believe our current approach of not applying a realization factor to the 40% of impaired loans that were re-underwritten will almost certainly result in a lower total subrogation recovery than if we had applied a realization factor to an extrapolation of the entire impaired loan population (had we believed such an extrapolation could be performed in a statistically valid manner). As such, we believe our current approach is an appropriately conservative estimation process.
To illustrate this, consider the following hypothetical example. Our subrogation recovery at December 31, 2009 under the adverse sample approach was approximately $450 million, which represented 40% of the impaired loan population. There was a 97% breach rate, thus the population of re-underwritten loans was $464m ($450m / 97%) and the total impaired loan population was $1,160m ($464m / 40%). Thus, if we were to assume the 40% of re-underwritten loanswere a statistically valid sample and we were able to extrapolate the 97% breach rate to the total impaired population, this would equate to
July 1, 2010
Page 6
$1,125m ($1,160m x 97%) of loan breaches. In order for the extrapolated subrogation recovery estimate to equate to the original subrogation recovery estimate of $450m (which did not have a realization factor applied), we would have to apply a realization factor of 40% ($450/$1,125), which is significantly lower than the realization factor we used in our random sample approach and one which we believe is very unrealistic. [Note that we previously disclosed the actual realization factor we used in our random sample approach in a confidential letter to the SEC dated February 8, 2010.] Also, the above example does not even consider breaches that may occur in unimpaired loans, which as mentioned above we believe is quite possible.
With respect to item (ii), it is unclear ifnot applying a realization factor to the re-underwritten population is statistically valid, per se. However it was not our intention to imply in our disclosures that subrogation recovery estimates should only be made in a statistically valid manner nor do we believe this is required by the accounting guidance. We believe the applicable accounting guidance allows for less precise measurement, such as the one we have applied, and this measurement is subject to the same cost-benefit constraints as discussed in our response to comment 2 above (Concepts Statement 7, paragraph 51). Therefore, with respect to item (iii), our subrogation estimate under our current adverse sample approach should not be considered a misstatement of our financial statements. Rather our approach was the best alternative available in light of our belief that we could not make a more precise measurement using a statistically valid extrapolation.
In the course of our discussions with the Staff, the question was raised whether we will continue to be reviewing impaired loan files under the adverse sample approach. Given the pervasiveness of the breaches noted above (97%) in the impaired loan files already reviewed, we have not significantly expanded the scope of our review since we do not currently believe the benefit of performing significant additional reviews warrants the added cost. Nonetheless, we are currently exploring alternative modeling techniques with the assistance of an outside consultant that may enable us to perform a statistically valid extrapolation on the impaired loan files we currently have access to. To the extent this more precise approach is feasible, given that this would likely involve a fairly lengthy vetting process both internally and with our external auditors; we would not anticipate implementing any changes to our modeling approach before the September 30, 2010 reporting period. Additionally, any changes to our modeling approach would be appropriately disclosed in the respective reporting period.
4. | Please refer to the table and footnotes on page 9 of your response. Please revise the disclosure to link the gross claim liability, subrogation recoveries and net claim liability to the tabular disclosure on the page 139 of your 2009 Form 10K. Also revise the disclosure to reconcile the loss reserve and the gross potential recoveries per page 139 to the balance sheet. |
Ambac response:
We will be revise the abovementioned disclosures as requested beginning with our June 30, 2010 Form 10-Q filing.
5. | We acknowledge your response to comment 4. Please address the following: |
• | Revise the rollforward to clearly delineate (1) the recoveries first recognized in the current period and (2) increases(decreases) in recoveries that are a re-estimation of the opening balance. For example the additional transactions reviewed and the additional loan files reviewed for transactions where the adverse sample approach is used would be a component of recoveries first recognized in the current year; |
• | Quantify in the rollforward the change resulting from applying the random sample approach to transactions where the adverse sample approach was previously used; and |
July 1, 2010
Page 7
• | Quantify in the rollforward the change resulting from additional loan files reviewed for transactions where the adverse sample approach is used. Also disclose the number of additional loan files that were reviewed. |
Ambac response:
We have revised the abovementioned disclosures as requested beginning with our March 31, 2010 Form 10-Q filing.
6. | Please revise the disclosure to link the amount of credits and net par outstanding that make up your subrogation recoverable to the tabular disclosure shown on page 63 of your 2009 Form 10K, which shows the number of credits and net par outstanding that make up your RMBS loss reserve. Also, disclose what you are netting in the net loss reserve. |
Ambac response:
We will be revise the abovementioned disclosures as requested beginning with our June 30, 2010 Form 10-Q filing.
7. | Please revise the disclosure to link the loss reserves and recoveries to the tabular disclosure shown on page 74 and 75 of your 2009 Form 10K. |
Ambac response:
We will be revise the abovementioned disclosures as requested beginning with our June 30, 2010 Form 10-Q filing.
Please feel free to contact me at (212) 208-3393 or Richard Alger at (212) 208-3196 should you require further information or have any questions.
Sincerely,
/s/ Robert Eisman |
Robert Eisman |
Senior Managing Director and |
Chief Accounting Officer |
Copy to: | David Wallis | |
President and Chief Executive Officer | ||
Ambac Financial Group, Inc. | ||
Kevin J. Doyle, Esq. | ||
Senior Vice President and General Counsel | ||
Ambac Financial Group, Inc. |
July 1, 2010
Page 8
David Trick | ||
Senior Managing Director and Chief Financial Officer | ||
Ambac Financial Group, Inc. | ||
Michael Groll, Esq. | ||
Dewey & LeBeouf LLP | ||
Richard Spitzer, Esq. | ||
Dewey & LeBeouf LLP | ||
Joel Parker | ||
Accounting Branch Chief | ||
Securities and Exchange Commission | ||
Paul Laurenzano | ||
Partner | ||
KPMG |
July 1, 2010
Page 9
Pro-Forma Disclosure black-lined from the March 31, 2010 Form 10-Q
Representation and Warranty Breaches by RMBS Transaction Sponsors:
In an effort to better understand the unprecedented levels of delinquencies, Ambac engaged consultants with significant mortgage lending experience to review the underwriting documentation for mortgage loans underlying certain insured RMBS transactions. These transactions which have exhibited exceptionally poor performance were chosen for further examination of the underwriting documentation supporting the underlying loans. Factors Ambac believes to be indicative of this poor performance include (i) increased levels of early payment defaults, (ii) the significant number of loan liquidations or charge-offs and resulting high level of losses, and (iii) the rapid elimination of credit protections inherent in the transactions’ structures. With respect to item (ii), “loan liquidations” refers to loans for which the servicer has liquidated the related collateral and the securitization has realized losses on the loan; “charge-offs” refers to loans which have been written off as uncollectible by the servicer, thereby generating no recoveries to the securitization, and may also refer to the unrecovered balance of liquidated loans. In either case, the servicer has taken such actions as it has deemed viable to recover against the collateral, and the securitization has incurred losses to the extent such actions did not fully repay the borrower’s obligations. Generally, the sponsor of the transaction provides representations and warranties with respect to the securitized loans including the loan characteristics, the absence of fraud or other misconduct in the origination process, including those attesting to the compliance of home loans with the prevailing underwriting policies. Per the transaction documents, the sponsor of the transaction is contractually obligated to repurchase, cure or substitute any loan that breaches the representations and warranties. Substitution is generally limited to two years from the closing of the transaction and the cure remedy is permitted only to the extent cure is possible.
Subsequent to the forensic exercise of examining loan files to ascertain whether the loans conformed to the representations and warranties, we submit nonconforming loans to the sponsor for repurchase. For all of the transactions reviewed by Ambac, the substitution remedy is no longer available. To effect a repurchase, depending on the transaction, the sponsor is contractually required to repurchase the loan at (a) for loans which have not been liquidated or charged off, either (i) the current unpaid principal balance of the loan, (ii) the current unpaid principal balance plus accrued unpaid interest, or (iii) the current unpaid principal balance plus accrued interest plus unreimbursed servicer advances/expenses and/or trustee expenses resulting from the breach of representations and warranties that trigger the repurchase, and (b) for a loan that has already been liquidated or charged-off, the amount of the realized loss. Notwithstanding the material breaches of representations and warranties, Ambac has continued to pay claims submitted under the financial guarantee insurance policies related to these securitizations. In cases where loans are repurchased by a sponsor, the effect is typically to offset current period losses and then to increase the over-collateralization of the securitization, depending on the extent of loan repurchases and the structure of the securitization. Specifically, the repurchase price is paid by the sponsor to the securitization trust which holds the loan. The cash becomes an asset of the trust, replacing the loan that was repurchased by the sponsor. On a monthly basis the cash received related to loan repurchases by the sponsor is aggregated with cash collections from the underlying mortgages and applied in accordance with the trust indenture payment waterfall. This payment waterfall typically includes principal and interest payments to the note holders, various expenses of the trust and reimbursements to Ambac, as financial guarantor, for claim payments made in previous months.With respect to transactions where Ambac has recorded estimated subrogation recoveries (as further described below), Ambac insures all or a portion of the senior tranches in the capital structure of the issuer, thus any sponsor cash received from loan repurchases would entirely benefit Ambac or Ambac-insured note holders. Notwithstanding the reimbursement of previous monthly claim payments, to the extent there continues to be insufficient cash in the waterfall in the current month to make scheduled principal and interest payments to the note holders, Ambac is required to make additional claim payments to cover this shortfall.
Ambac’s estimate of subrogation recoveries includes two components: (1) estimated dollar amounts of loans with material breaches of representations and warranties based on an extrapolation
July 1, 2010
Page 10
of the breach rate identified in a random sample of loans taken from the entire population of loans in a securitization (“random sample approach”) and (2) dollar amount of actual loans with identified material breaches of representations and warranties discovered from samples of impaired loans in a securitization (“adverse sample approach”). We do not include estimates of damages in our estimate of subrogation recoveries under either approach. The amount the sponsors believe to be their liability for these breaches is not known.
The random sample approach to estimate subrogation recoveries was based on obtaining a statistically valid random sample for all the original loans in the pool. First, a “breach rate” was computed by dividing (i) the loans identified in sample as having breached representations and warranties by (ii) the total sample size. Second, an extrapolation to the entire loan pool was performed by multiplying the breach rate by the sum of (a) the current unpaid loan pool balance (“CULPB”) plus (b) realized losses resulting from loan liquidations or charge-offs to date, to compute an estimated repurchase obligation. The CULPB includes principal only on non-charged-off and non-liquidated loans, and the realized losses include principal, interest and unreimbursed servicer advances and/or trustee expenses on charged-off and liquidated loans. As a result, the CULPB and realized loss components, which are used in extrapolating the estimated repurchase obligation, do not precisely correspond to each sponsor’s contractual repurchase obligation as defined in the transaction documents. Nonetheless, the CULPB and realized loss components are provided through regular trustee reports we receive in the normal course of our surveillance of these transactions and is the best information we have available to estimate the sponsor’s repurchase obligation under the random sample approach. Third, a realization factor was then applied to the estimated repurchase obligation to compute the undiscounted subrogation recovery, which incorporates Ambac’s views about the uncertainties surrounding the settlement negotiation and litigation processes. The realization factor was developed from a range of realization factors using Ambac’s own assumptions about the likelihood of outcomes based on all the information available to it including (i) discussions with external legal counsel and their views on ultimate settlement, (ii) recent experience with loan put back negotiations where the existence of a material breach was debated and negotiated at the loan level, and (iii) the pervasiveness of the breach rates. Finally, a discount factor was applied to the undiscounted subrogation recovery to compute the estimated subrogation recovery using the assumptions discussed in the paragraph subsequent to the next table below.
The subrogation recovery estimate we have recorded produced by the above-described random sample approach includes all breached loans which we believe the sponsor is contractually required to repurchase, including extrapolation to a loan pool which includes loans which we do not anticipate defaulting (i.e. performing loans). In theory, a performing loan should have little or no effect on Ambac’s anticipated claim payments, regardless of whether or not the sponsor repurchases the loan. In other words, since there will be sufficient cash flows to service the notes in either situation (i.e. whether cash is received from a sponsor loan repurchase or whether cash is received from the underlying performing loan), there should be no claim payment under Ambac’s insurance policy regardless of the outcome. Nonetheless, we have recorded a subrogation recovery for certain performing loans because we believe the breaches of representations and warranties are so pervasive that a judicial court would deem it impractical to have the sponsor re-underwrite every loan in a given transaction and repurchase individual loans that have breached. Rather we believe there is established precedent for the utilization of a statistical sampling and extrapolation methodology across a population to prove liability and damages where it would be impractical to make a determination on an individual basis. We believe the court would likely require a monetary settlement based on a reasonable methodology, such as our random sample approach, and limit such a settlement to Ambac’s incurred losses. That is, the settlement monies would be either remitted directly to Ambac, placed in the securitization trust, or otherwise held under an arrangement for the benefit of the securitization trust, however individual loans would not be repurchased from the trust. In either case, those settlement monies would offset past and future Ambac claim payments. Consequently, since the sponsor is contractually obligated to repurchase those loans which breach representations and warranties regardless of whether they are current or defaulted, we believe the appropriate measure in estimating subrogation recoveries is to apply the breach rate to both performing and defaulted loans.
July 1, 2010
Page 11
The adverse sample approach to estimate subrogation recoveries was based on a sample taken from those loans in the pool that were impaired, meaning loans greater than 90 days past due, charged-off, in foreclosure, REO or bankruptcy. The estimated subrogation recovery under this approach represents 100% of the original principal balance of those specific loans identified as having not met the underwriting criteria or otherwise breaching representations and warranties (i.e. the adverse loans), multiplied by a discount factor using the same assumptions used for the discount factor in the random sample approach. For transactions subject to the adverse sample approach, given Ambac’s limitations in developing a statistically valid random sample and its belief that the subrogation estimate under this approach is inherently conservative (for reasons discussed below), Ambac did not attempt to develop probability-weighted alternative cash flow scenarios as it believes such results would not be meaningful. The three primary differences between this adverse sample approach and the random sample approach, discussed in the previous paragraph, are as follows:
(i) | There is no extrapolation to the CULPB and realized losses under the adverse sample approach. At March 31, 2010, the adverse sample approach continues to be used for ten transactions that are with the same sponsor, who has limited our access to the underlying loan files and, therefore, a statistically valid random sample from the entire loan pool cannot be selected. This is in contrast to the transactions subject to the random sample approach where Ambac’s access to individual loan files has not been limited and the Company, therefore, has been able to develop a statistically valid representative sample. |
(ii) | The adverse sample approach is only based on the original principal balance rather than the principal balance at the time of default and liquidation or charge-off. Furthermore, it does not include other components of the sponsor’s contractual repurchase obligation where the sponsor is also obligated to repay accrued interest, servicer advances and/or trustee expenses. The adverse sample approach relies on individual loan level data where all of the components of the sponsor’s buyback obligation have not been specifically provided by the sponsor nor is easily estimable. For example, home equity lines of credit (HELOCs) are revolving loans whose principal balances may be higher or lower at the time of default and liquidation or charge-off than at the time of origination. However, given the limited information available to Ambac in estimating such principal balances at the time of liquidation or charge-off, the original principal balance must be used in calculating subrogation recoveries. Another example is closed-end second lien RMBS where the interest due on a particular loan will be a function of the length of delinquency prior to liquidation or charge-off, and cannot be readily estimated. Incremental costs, including fees and servicer advances for such items as property taxes and maintenance, are likewise not readily estimated. |
(iii) | Unlike the random sample approach, for the adverse sample approach Ambac did not apply a realization factor to the estimated repurchase obligation for the adverse loans related to uncertainties surrounding settlement negotiation or litigation processes given that the adverse loans selected represent only 40% of the impaired population of loans, only 4% of the original number of loans in the pool and the breach rate in the sample was pervasive. In other words, because the adverse sample size represents only a fraction of the population of impaired loans and a very small proportion of the original loans in the pools, Ambac believes there is an ample population of additional impaired loans where breaches of representations and warranties exist that could potentially replace any adverse loans it already identified that might be successfully challenged in negotiations or litigation. |
During the latter half of 2009, Ambac expanded its use of the random sample approach for estimating the amount of subrogation recoveries to include all transactions where a statistically valid random sample of loan files was available. Given the scale of the losses in the RMBS portfolio, and the evidence of pervasive breaches, Ambac believes limiting remediation credit to the loan amounts where actual breaches have been discovered (i.e. the adverse sample approach) is inconsistent with its
July 1, 2010
Page 12
gross claim projection methodology and understates the amount Ambac is expected to recover from sponsors. The adverse sample approach continues to be used only on transactions insured for one sponsor (10 different transactions) who has limited Ambac’s access to the underlying loan files and therefore a statistically valid random sample from the entire loan pool cannot be selected.
Ambac has updated its estimated subrogation recoveries from $2,026.3 million at December 31, 2009 to $2,069.2 million at March 31, 2010. The balance of subrogation recoveries and the related claim liabilities at March 31, 2010 and December 31, 2009 are as follows:
($ in millions)
March 31, 2010 | December 31, 2009 | ||||||||||||||||||||||||||||
Method | Count | Claim liability | Subrogation recoveries(1) | Claim liab. net of subrogation recoveries | Method | Count | Claim liability | Subrogation recoveries | Claim liab. net of subrogation recoveries | ||||||||||||||||||||
Adverse Samples | 10 | (2) | $ | 884.5 | ($ | 486.2 | ) | ($ | 398.3 | ) | Adverse Samples | 10 | $ | 749.4 | ($ | 450.2 | ) | $ | 299.2 | ||||||||||
Random Samples | 10 | (3) | 827.6 | (1,583.0 | ) | (755.4 | ) | Random Samples | 9 | 931.5 | (1,576.1 | ) | (644.6 | ) | |||||||||||||||
Totals | 20 | $ | 1,712.1 | ($ | 2,069.2 | ) | ($ | 357.1 | ) | Totals | 19 | $ | 1,680.9 | ($ | 2,026.3 | ) | ($ | 345.4 | ) |
(1) | The amount of recorded subrogation recoveries related to each securitization is limited to ever-to-date paid losses plus the present value of projected future paid losses for each policy. To the extent significant losses have been paid but not yet recovered, the recorded amount of subrogation recoveries may exceed the projected future paid losses for a given policy. The net cash inflow for these policies is recorded as a “Subrogation recoverable” asset. For those transactions where the subrogation recovery is less than projected future paid losses, the net cash outflow for these policies is recorded as a “Loss and loss expense reserve” liability. Of the $2,069.2 million of subrogation recoveries recorded at March 31, 2010, $1,750.4 million was included in “Subrogation recoverable” and $318.8 million was included in “Loss and loss expense reserves.” |
(2) | Of these 10 transactions, 7 contractually require the sponsor to repurchase loans at the unpaid principal balance and 3 contractually require the sponsor to repurchase loans at unpaid principal plus accrued interest. However, for reasons discussed above in the description of the adverse sample approach, our estimated subrogation recovery for these transactions may not include all the components of the sponsor’s contractual repurchase obligation. |
(3) | Of these 10 transactions, 3 contractually require the sponsor to repurchase loans at unpaid principal plus accrued interest and 7 contractually require the sponsor to repurchase loans at unpaid principal plus accrued interest plus servicer advances/expense and/or trustee expenses. However, for reasons discussed above in the description of the random sample approach, our estimated subrogation recovery for these transactions may not include all the components of the sponsor’s contractual repurchase obligation. |
While the obligation by sponsors to repurchase loans with material breaches is clear, generally the sponsors have not yet honored those obligations. Ambac’s approach to resolving these disputes has included negotiating with individual sponsors at the transaction level and in some cases at the individual loan level and has resulted in the repurchase of some loans. Ambac has utilized the results of the above described loan file examinations to make demands for loan repurchases from
July 1, 2010
Page 13
sponsors or their successors and, in certain instances, as a part of the basis for litigation filings. Ambac has initiated and will continue to initiate lawsuits seeking compliance with the repurchase obligations in the securitization documents. Ambac estimates that it will take approximately three years from the initiation of litigation with the sponsor to ultimate resolution. Based on this estimate as a basis for projecting the future subrogation cash flows, Ambac assumes, on average, approximately three and a half years to collect recoveries, discounted at a risk-free rate of 2.50%. Estimated recoveries will continue to be revised and supplemented as the scrutiny of the mortgage loan pools progresses.
We have performed the above-mentioned, detailed examinations on a variety of second-lien transactions and three first-lien transactions that have experienced exceptionally poor performance. However, the loan file examinations and related estimated recoveries we have reviewed and recorded to date have been limited to only those transactions whose sponsors (or their successors) are subsidiaries of large financial institutions, all of which carry an investment grade rating from at least one nationally recognized rating agency. A total of six sponsors represent the twenty transactions which have been reviewed as of March 31, 2010. Each of these financial institutions has significant financial resources and an ongoing interest in mortgage finance. Additionally, in the case of successor institutions, we are not aware of any provisions that explicitly preclude or limit the successors’ obligations to honor the obligations of the original sponsor. As a result, we did not make any significant adjustments to our estimated subrogation recoveries with respect to the credit risk of these sponsors (or their successors). We believe that focusing our loan remediation efforts on large financial institutions first will provide the greatest economic benefit to Ambac. Ambac retains the right to review all RMBS transactions for representations and warranties breaches. Since a significant number of other second-lien and first-lien transactions are also experiencing poor performance, management is considering expanding the scope of this effort.