 | One Astoria Federal Plaza Lake Success, NY 11042-1085 (516) 327-3000 |
November 25, 2008
Mr. Kevin W. Vaughn
Branch Chief
Division of Corporation Finance
United States Securities and Exchange Commission
450 Fifth Street, N.W.
Washington, D.C. 20549
Mail Stop 4561
| Re: | Astoria Financial Corporation |
| Form 10-K for Fiscal Year Ended December 31, 2007 | |
| Form 10-Q for Fiscal Quarter Ended March 31, 2008 | |
| Form 10-Q for Fiscal Quarter Ended June 30, 2008 | |
| File No. 001-11967 |
Dear Mr. Vaughn:
We wish to acknowledge receipt of your letter dated November 3, 2008 concerning the Securities and Exchange Commission’s review of the disclosures contained in Astoria Financial Corporation’s (the “Company”) Form 10-K for the Fiscal Year Ended December 31, 2007, Form 10-Q for the Fiscal Quarter Ended March 31, 2008 and Form 10-Q for the Fiscal Quarter Ended June 30, 2008. The following sets forth the Company’s responses to your comments.
Form 10-K for the year ended December 31, 2007
Business
Lending Activities: One-to-Four Family Mortgage Lending, page 4
1. | We note your disclosures here and throughout your document regarding your interest only Hybrid ARM loans. We also note that during 2006 you began underwriting these loans at the fully indexed rate. Please tell us and revise future filings to clarify if that policy change affected all interest only products, including interest only ARM loans and interest only Hybrid ARM loans. |
Our only interest-only one-to-four family ARM loan product is our interest-only hybrid ARM loan, which has an interest only period of ten years after which principal and interest payments are required to amortize the loan over the remaining loan term. We do not originate one year ARM loans. The ARM loans in our portfolio which currently reprice annually represent hybrid ARM loans (interest-only and amortizing) which have passed their initial fixed rate period. Our interest-
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 2
only and amortizing hybrid ARM loans may be offered with an initial interest rate which is less than the fully indexed rate for the loan at the time of origination (discounted rate). We determine the initial interest rate in accordance with market and competitive factors giving consideration to the spread over our funding sources in conjunction with our overall interest rate risk management strategies. We do not originate negative amortization loans, payment option loans or other loans with short-term interest-only periods. Prior to 2006, we would underwrite our interest-only hybrid ARM loans using the initial note rate, which may have been a discounted rate. In 2006, we began underwriting our interest-only hybrid ARM loans based on a fully amortizing loan (in effect underwriting interest-only hybrid ARM loans as if they were amortizing hybrid ARM loans). Additionally, in 2007, we began underwriting our interest-only hybrid ARM loans at the higher of the fully indexed rate or the initial note rate.
We will, in future filings, clarify that beginning in 2007, all of our interest-only one-to-four family ARM loan originations (which are all interest-only hybrid ARM loans) are underwritten based on a fully amortizing loan using the higher of the fully indexed rate or the initial note rate.
2. | Considering the different credit risks associated with the underwriting requirements of your ARM products, please provide us and revise future filings here and elsewhere in your document, as appropriate, to include disaggregated information of your interest only loans to disclose the amounts and percentages of these loans that are underwritten at the fully indexed rate and those that are not. It may be helpful to expand the table provided in the Asset Quality section of MD&A to include this disclosure. |
One-to-four family interest-only hybrid ARM loans totaled $7.65 billion as of December 31, 2007, of which $2.31 billion, or 30%, were underwritten at the fully indexed rate and $5.34 billion, or 70%, were underwritten at the initial note rate, which may have been a discounted rate. One-to-four family interest-only hybrid ARM loans totaled $7.51 billion as of September 30, 2008, of which $2.93 billion, or 39%, were underwritten at the fully indexed rate and $4.58 billion, or 61%, were underwritten at the initial note rate, which may have been a discounted rate.
We will, in future filings, include the balances of interest-only hybrid ARM loans that were underwritten at the fully indexed rate and those that were underwritten at the initial note rate, which may have been a discounted rate, as a footnote to the tables on pages 67 and 103 of our 2007 Annual Report on Form 10-K which detail the composition of our total one-to-four family loan portfolio, and elsewhere in our document, as appropriate.
3. | We note that at December 31, 2007 your portfolio of reduced documentation loans consists primarily of SIFA loans. Please tell us and consider disclosing in future filings the amount of SISA and Super Streamline loans held in your portfolio. Consider expanding your tabular breakdowns at various locations in your filing similar to that on page 103 to separately quantify the SISA and Super Streamline loans. |
We discontinued originating SISA and Super Streamline loans in the 2007 second quarter and SIFA loans in the 2007 fourth quarter. At December 31, 2007, SISA loans totaled $412.3 million, or 4% of our total one-to-four family mortgage loan portfolio, and Super Streamline loans totaled $47.3
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 3
million, or 0.4% of our total one-to-four family mortgage loan portfolio. At September 30, 2008, SISA loans totaled $367.2 million, or 3% of our total one-to-four family mortgage loan portfolio, and Super Streamline loans totaled $38.2 million, or 0.3% of our total one-to-four family mortgage loan portfolio.
We will, in future filings, include as a footnote to the tables on pages 67 and 103 of our 2007 Annual Report on Form 10-K which detail the composition of our total one-to-four family loan portfolio, the balances of SISA and Super Streamline loans as of the dates presented in the tables until such time that they become clearly immaterial.
4. | At various places in your document you refer to your sound underwriting policies and strong risk management systems utilized to prudently manage the greater credit risk posed by ARM and interest-only ARM loans. Please tell us and revise future filings to include an expanded discussion of these policies and systems. Please include a discussion of how you monitor credit risk on loans with future interest reset dates and loans that were not underwritten at fully indexed rates. Clearly identify any changes to your underwriting policies or risk management systems implemented during the periods presented. Consider the need to provide an expanded discussion of your risk management policies for these loans and their affects on the allowance for loan and lease losses within the Asset Quality section of your document. |
Our underwriting policies and risk management systems include a variety of factors and analyses. These include, but are not limited to, the determination of the markets in which we lend; the products we offer and the pricing of those products; the evaluation of potential borrowers and the characteristics of the property supporting the loan; the monitoring and analyses of the performance of our portfolio, in the aggregate and by segment, at various points in time and trends over time; and our collection efforts and marketing of non-performing loans and foreclosed properties.
The objective of our one-to-four family mortgage loan underwriting is to determine whether timely repayment of the debt can be expected and whether the property that secures the loan provides sufficient value to recover our investment in the event of a loan default. We review each loan individually utilizing such documents as the loan application, credit report, verification forms, tax returns and any other documents relevant and necessary to qualify the potential borrower for the loan. We analyze the credit and income profiles of potential borrowers and evaluate all aspects of the potential borrower’s credit history including credit (FICO) scores. We do not base our underwriting decisions solely on FICO scores. We consider the potential borrower’s income, history of debt management and net worth. We perform income and debt ratio analyses as part of the credit underwriting process. Additionally, we obtain independent appraisals to establish collateral values to determine loan-to-value (LTV) ratios.
We monitor our market areas and the performance and pricing of our various loan product offerings to determine the prudence of continuing to offer such loans and to determine what changes, if any, should be made to our product offerings and related underwriting. We offer amortizing hybrid ARM loans and interest-only hybrid ARM loans. In 2006 we began underwriting our interest-only hybrid ARM loans based on a fully amortizing loan and in 2007 we began underwriting such loans at the
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 4
higher of the fully indexed rate or the initial note rate. During 2008, we discontinued originating interest-only hybrid ARM loans with an initial fixed rate period of three years and currently only originate such loans with initial fixed rate periods of five and seven years. Since our interest-only hybrid ARM loans have a relatively long period to the principal payment adjustment, we believe there is minimal additional credit risk compared to an amortizing loan due to the longer period for the borrower’s income to adjust to anticipated higher future payments. We do not originate negative amortization loans, payment option loans or other loans with short-term interest-only periods.
We have reduced documentation loans in our one-to-four family mortgage loan portfolio. Reduced documentation loans are comprised primarily of SIFA (stated income, full assets) loans which require a potential borrower to complete a standard mortgage loan application and require the verification of a potential borrower’s asset information on the loan application, but not the income information provided. In addition, SIFA loans require the receipt of an appraisal of the real estate used as collateral for the mortgage loan and a credit report on the prospective borrower. To a lesser extent, our reduced documentation loans also include SISA (stated income, stated assets) loans and Super Streamline loans. We discontinued originating SISA and Super Streamline loans in the 2007 second quarter and discontinued originating SIFA loans in the 2007 fourth quarter. As a result, effective January 2008, we no longer offer reduced documentation loans.
As a community bank, our risk management systems are comprised of the analyses we use to monitor our loan portfolio. We segment our one-to-four family loan portfolio by interest-only and amortizing loans and by state. We monitor our loss experience and delinquency levels and trends for these segments of our portfolio. We monitor credit risk on interest-only hybrid ARM loans that were underwritten at the initial note rate, which may have been a discounted rate, in the same manner as we monitor credit risk on all interest-only hybrid ARM loans. To recognize the additional risks associated with interest-only hybrid ARM loans, we apply a higher reserve coverage percentage to that portion of the portfolio in determining our allowance for loan losses. We monitor interest rate reset dates of our portfolio, in the aggregate, and the current interest rate environment and consider the impact, if any, on the borrowers’ ability to continue to make timely principal and interest payments in determining our allowance for loan losses. During the fourth quarter of 2007, we began experiencing more significant increases in non-performing loans and net loan charge-offs primarily as a result of the decline in the housing and real estate markets, as well as the overall economic environment. As a result, during 2008 we began further segmenting the composition of our one-to-four family portfolio by full documentation and reduced documentation and by year of origination to perform additional analyses of our loss experience, delinquency levels and trends for these portfolio segments. We also monitor property value trends in our market areas to determine what impact, if any, such trends may have on our loan-to-value ratios and the adequacy of the allowance for loan losses.
We will, in future filings, expand our discussion to include the above information regarding our underwriting policies and risk management systems and continue to clearly identify any changes implemented during the periods presented.
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 5
Construction Loans, page 6
5. | Please tell us and revise future filings to disclose whether or not you maintain interest reserves on your construction loans. If applicable, please address the following: |
| · | If you do require interest reserves, please quantify the amount of such loans and accompanying reserves at December 31, 2007 and as of the balance sheet date for each quarter in calendar year 2008. |
| · | Describe how you monitor such projects throughout their lives to make sure the properties are moving along as planned such that it is appropriate to continue to capitalize the interest to the loan. |
| · | Discuss whether you have ever extended, renewed or restructured terms of the related loans, and the reasons for the changes. |
Our construction loan portfolio totaled $77.7 million, or 0.48% of our total loan portfolio, at December 31, 2007 and $58.1 million, or 0.35% of our total loan portfolio, at September 30, 2008. Substantially all of our construction loans are multi-family construction projects. We are not currently originating construction loans.
Construction loans are funded, at the request of the borrower, not more than once per month, based on the work completed, and are monitored by an independent professional construction engineer and our commercial real estate lending department throughout the life of the project. Within each construction loan, there is a budget for “hard costs” (actual work in place and construction materials) and “soft costs” (interest reserves and architectural fees). An interest reserve amount is established as part of the soft cost budget when the loan is approved. The amount of interest is a function of the loan amount, term and interest rate. Interest is advanced to the borrower, upon request, in relative proportion to the percentage of completion of the construction project. The amount of interest advanced is added to the total outstanding principal under the construction loan commitment. If the project is not progressing as scheduled, additional interest is not advanced and the borrower is required to fund the interest payments until the progress of the project is back on schedule. Once the interest reserve amount established at the origination of the loan is fully funded, the borrower is required to fund the interest payments.
At December 31, 2007, construction loans totaled $77.7 million, of which $4.7 million represented interest reserves. At March 31, 2008, construction loans totaled $67.9 million, of which $4.6 million represented interest reserves. At June 30, 2008, construction loans totaled $60.3 million, of which $4.0 million represented interest reserves. At September 30, 2008, construction loans totaled $58.1 million of which $3.4 million represented interest reserves.
Construction loans are reviewed for extensions upon expiration of the loan term. Extensions may be granted to allow for the completion of the project, marketing or sales of units or to provide permanent financing. Provided that the loan is performing satisfactorily, the request for an extension is typically granted for a period of 6 months at a time. We had no restructured construction loans, as defined under SFAS No. 15 and SFAS No. 114, in our portfolio at the noted reporting dates.
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 6
We do not believe it is necessary to expand our disclosures related to construction loans due to the immateriality of our interest reserve balances and the size of our construction loan portfolio.
Management’s Discussion and Analysis of Financial Condition and Result of Operation
Executive Summary, page 41
6. | We note your discussion that you are not exposed to sub-prime lending, which you say has been the riskier sector of the residential housing market. Please consider the need to balance this discussion with one that addresses the specific risks associated with interest-only loan products to include those products that were not written at the fully indexed interest rate. |
Generally, ARM loans pose credit risks somewhat greater than the risks posed by fixed rate loans primarily because, as interest rates rise, the underlying payments of the borrower increase when the loan is beyond its initial fixed rate period, particularly if the interest rate during the initial fixed rate period was at a discounted rate, increasing the potential for default. Interest-only ARM loans have an additional potential risk element when the loan payments adjust after the tenth anniversary of the loan to include principal payments, resulting in a further increase in the underlying payments.
In future filings, we will enhance our discussion of the risks associated with the different one-to-four family loans types in our portfolio to address our interest-only hybrid ARM loans which were underwritten at the initial note rate, which may have been a discounted rate. We believe the best location for this discussion is the “Lending Activities” section of Part I, Item 1. “Business” in the Annual Report on Form 10-K, rather than as part of the “Executive Summary” in Part II, Item 7. “Management’s Discussion and Analysis of Financial Condition and Results of Operations” (MD&A). In our Executive Summary, and in other locations in our filings, where we reference that we have never actively participated in high-risk subprime residential lending, we will add a cross reference to the appropriate section in our filings which discusses the risks associated with our different loan products. We will continue to balance our discussion that we have never actively participated in high-risk subprime residential lending with our discussion that we have been affected by the negative consequences arising from overall economic weakness and, in particular, a sharp downturn in the housing industry nationally, as well as economic and housing industry weaknesses in the New York metropolitan area specifically.
7. | We note your disclosure of the other-than-temporary impairment taken on two issues of FHLMC perpetual preferred securities. We also note that you are recognizing a further impairment of these assets in the quarter ended September 30, 2008, as reported on a Form 8-K filed on October 16, 2008. Please tell us and revise future filings to disclose how you identified and measured the impairment at December 31, 2007 and at September 30, 2008. Also, discuss the facts and circumstances that changed between the periods that resulted in the recognition of a further impairment charge and how they resulted in impairment recognition in the third quarter rather than in an earlier period. |
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 7
During the 2007 fourth quarter, we recorded a $20.5 million other-than-temporary impairment (OTTI) charge to reduce the carrying amount of our investment in two issues of Freddie Mac perpetual preferred securities to the securities’ market values of $83.0 million at December 31, 2007. The decision to recognize the OTTI charge was based on the consistent application of our OTTI accounting policy which is based on SAB No. 59. We concluded that the severity of the decline in the market value of these securities, caused by Freddie Mac’s announcement of negative financial results and capital raising activity during the fourth quarter of 2007, and the unlikelihood of any near-term market value recovery met our OTTI policy requirements. This information was disclosed on page 46 of our 2007 Annual Report on Form 10-K in “Critical Accounting Policies – Securities Impairment.”
At March 31, 2008 and June 30, 2008, we evaluated our Freddie Mac securities as part of our ongoing OTTI evaluation process to determine if the decline in the fair value of these securities below their cost basis was other-than-temporary and determined, at those times, that the decline in the fair value was not other-than-temporary. These determinations were based on our evaluation of the duration and magnitude of the impairment at March 31, 2008 and June 30, 2008, our intent and ability to hold the securities, our assessment of the reasons for the decline in value and the likelihood of a near-term recovery.
The two issues of Freddie Mac securities we hold are the Series H (2,000,000 shares) and the Series L (400,000 shares). The duration of the unrealized losses was considered short for both securities as of March 31, 2008 and June 30, 2008. The Series H had traded above its carrying amount on March 7, 2008 and June 23, 2008 and the Series L had traded above its carrying amount on February 27, 2008. Additionally, the magnitude of the unrealized losses was considered low at March 31, 2008 and June 30, 2008. The aggregate impairment at March 31, 2008 and June 30, 2008, as measured by the difference between the carrying amount and market value of the securities as of those dates, totaled $5.1 million, or only 6% of the securities’ carrying amounts, at March 31, 2008 and totaled $6.1 million, or only 7% of the securities’ carrying amounts, at June 30, 2008. As of March 31, 2008 and June 30, 2008, we had the intent and ability to hold the securities, which had been our position since the securities were purchased in 1998 and 1999. As of March 31, 2008 and June 30, 2008, the yields on our Freddie Mac securities were attractive and, prior to the suspension of the dividends by the Federal Housing Finance Agency, or FHFA, on September 7, 2008, neither of these securities had experienced an interruption in dividend payments over their ten year lives.
We concluded at June 30, 2008 that the likelihood of a near-term recovery of the value of these securities could be expected based on several factors: (1) Freddie Mac, as a government sponsored enterprise (GSE), had always had the implied backing of the U. S. Government, and we believed that any government intervention would more likely than not further strengthen that implied guarantee; (2) the Director of the Office of Federal Housing Enterprise Oversight had made public statements in support of Freddie Mac’s role and that Freddie Mac was adequately capitalized; (3) the spread between Agency issues and U. S. Treasuries was at historically wide levels and a return to more normal levels would require an increase in the price of Agency paper; and (4) the median earnings forecasts for Freddie Mac at that time showed a pattern of continued improvement to earnings.
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 8
Subsequent to June 30, 2008, Freddie Mac was subject to significant market disruption, as well as political and regulatory discussions, which caused its preferred stock fair values to decline further. We disclosed these subsequent events on page 51 of our June 30, 2008 Quarterly Report on Form 10-Q in “Risk Factors.” In addition, we disclosed that we would continue to monitor the fair value of the Freddie Mac securities we hold as related market events unfolded as part of our ongoing OTTI evaluation process and that no assurance could be given that we would not need to recognize an OTTI charge related to these securities if the fair values did not recover in the future.
During the 2008 third quarter, we recorded a $77.7 million OTTI charge to again reduce the carrying amount of our Freddie Mac securities to the securities’ market values of $5.3 million at September 30, 2008. The decision to recognize the OTTI charge was again based on the consistent application of our OTTI accounting policy, the decline in the market value of these securities in the 2008 third quarter and the unlikelihood of any near-term market value recovery. The significant decline in the market value occurred primarily as a result of the recently reported financial difficulties of Freddie Mac and the subsequent announcement, on September 7, 2008, by the U.S. Department of Treasury and the FHFA that, among other things, Freddie Mac was being placed under conservatorship; that the FHFA was assuming the powers of Freddie Mac’s Board and management; and that dividends on Freddie Mac preferred stock were suspended indefinitely. This information was disclosed on pages 23 and 24 of our September 30, 2008 Quarterly Report on Form 10-Q in “Critical Accounting Policies – Securities Impairment.”
We believe our previous disclosures related to our OTTI charges were complete and addressed the facts and circumstances that led to such charges for the year ended December 31, 2007 and the nine months ended September 30, 2008. Additionally, we believe our previous disclosures related to our OTTI evaluations, which resulted in no OTTI charges, for the three months ended March 31, 2008 and the six months ended June 30, 2008 were also complete. We will include the text in the sixth paragraph of this response in the “Executive Summary” and in “Critical Accounting Policies – Securities Impairment” in Part II, Item 7. “MD&A” of our 2008 Annual Report on Form 10-K. Additionally, we will include the text in the first and sixth paragraphs of this response in our securities footnote in Part II, Item 8. “Financial Statements and Supplementary Data.”
Allowance for Loan Losses, page 43
8. | At the top of page 44 you discuss the fact that you “segment (y)our loan portfolio into like categories, by composition and size and perform analyses against each category.” In future filings, please expand this section to more clearly discuss the extent to which your various types of loan products, specifically your types of interest-only loans and various levels of reduced documentation loans are segmented for credit analysis purposes. |
We segment our loan portfolio into like categories by composition and size and perform analyses against each category. These analyses include historical loss experience and delinquency levels and trends. We analyze our historical loss experience by category (loan type) over 3, 5, 10, 12 and 16 year periods. Losses within each loan category are stress tested by applying the highest level of charge-offs and the lowest amount of recoveries as a percentage of the average portfolio balance during those respective time horizons. In addition, we further segment the composition of our one-
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 9
to-four family portfolio by interest-only and amortizing loans and by state, and we segment our consumer and other loan portfolio by home equity lines of credit, business loans, revolving credit lines and installment loans and perform similar analyses against these portfolio segments. During the fourth quarter of 2007, we began experiencing more significant increases in non-performing loans and net loan charge-offs primarily as a result of the decline in the housing and real estate markets, as well as the overall economic environment. As a result, during 2008 we have further segmented the composition of our one-to-four family portfolio by full documentation and reduced documentation loans and by year of origination, and began performing additional analyses of our loss experience and delinquency levels and trends for these portfolio segments.
We will, in future filings, expand our disclosures to more fully describe the extent to which we segment our various types of loan products for credit analysis purposes.
Results of Operations
Provision for Loan Losses, page 57
9. | We note your extensive discussions here and in the Critical Accounting Policy section of how management develops and estimates the allowance for loan losses. We also note the significant deterioration of your asset quality ratios as set forth in the Selected Financial Data Section and the continuing deterioration of these ratios in the first and second quarters of calendar year 2008, which have resulted in exponentially larger provisions in each of those quarters. Please revise your disclosures in future interim and annual filings to more clearly disclose how management specifically considered the deteriorating asset quality ratios in developing and estimating the allowance for loan losses and the provision at December 31, 2007 and at the balance sheet date for each quarter in 2008. Please include a discussion of the material assumptions relied on by management which may not be apparent in your disclosures or reflected in your asset quality ratios and activity in the allowance for loan losses. Please provide us with your proposed disclosures. Some of the material trends that we note that appear to warrant further discussion are as follows: |
| · | We note that at December 31, 2007 the allowance for loan losses to total loans decreased approximately 7% while non-performing loans to total loans increased approximately 45% and non-performing loans to total loans increased approximately 65%. |
| · | We note that at March 31, 2008 the allowance for loan losses to non-performing loans decreased approximately 35% from December 31, 2007 while non-performing loans to total loans increased approximately 61%. |
| · | We note that at June 30, 2008 the allowance for loan losses to non-performing loans decreased 45% from December 31, 2007 and approximately 15% from March 31, 2008 while non-performing loans to total loans increased approximately 88% from December 31, 2007 and approximately 16% from March 31, 2008. |
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 10
| · | We note that charge-offs at June 30, 2008 increased approximately 114% from December 31, 2007 while the allowance for loan losses to total loans increased approximately 4%. |
Our allowance for loan losses is established and maintained through a provision for loan losses based on our evaluation of the probable inherent losses in our loan portfolio in accordance with SFAS No. 114 and SFAS No. 5. The allowance is comprised of both specific valuation allowances and general valuation allowances. Specific valuation allowances are established for our collateral dependent loans based on individual loan reviews. General valuation allowances represent loss allowances that have been established to recognize the inherent risks associated with our lending activities, but which, unlike specific valuation allowances, have not been allocated to particular problem loans.
Our determination of the allowance for loan losses includes a review of our historical loss experience and delinquency levels and trends, by loan type and segment, which are an integral part of our judgment in developing estimated loss percentages for our portfolio. In conjunction with a variety of other factors, we determine a range of estimated losses and appropriate allowance coverage percentages for each of our portfolio segments. Other factors include, but are not limited to, the size, composition, risk profile, delinquency levels and cure rates of our portfolio as well as our credit administration and asset management philosophies and procedures. We monitor property value trends in our market areas by reference to various industry and market reports, economic releases and surveys, and our general and specific knowledge of the real estate markets in which we lend, in order to determine what impact, if any, such trends may have on the level of our general valuation allowances. In determining our allowance coverage percentages for non-performing loans, we consider our historical loss experience with respect to the ultimate disposition of the underlying collateral. In addition, we evaluate and consider the impact that existing and projected economic and market conditions may have on the portfolio, as well as known and inherent risks in the portfolio. We also evaluate and consider our asset quality ratios as well as the allowance ratios and coverage percentages set forth in both peer group and regulatory agency data and any comments from the OTS resulting from their review of our general valuation allowance methodology during regulatory examinations. Our focus, however, is primarily on our historical loss experience, the composition and direction of loan delinquencies and the impact of current economic conditions. After evaluating these variables, we determine appropriate allowance coverage percentages for each of our portfolio segments and the appropriate level of our allowance for loan losses. Our allowance for loan losses and the related increases in the provision for loan losses for the year ended December 31, 2007 and the periods ended March 31, 2008, June 30, 2008 and September 30, 2008 reflect the increases in and composition of our loan delinquencies, non-performing loans, net loan charge-offs and overall loan portfolio, as well as our evaluation of the continued deterioration of the housing and real estate markets and the weakness in the overall economy.
At December 31, 2007, our allowance for loan losses totaled $78.9 million. During the nine months ended September 30, 2008, our provision for loan losses totaled $24.0 million and net loan charge-offs totaled $16.6 million which resulted in an allowance for loan losses of $86.3 million at September 30, 2008. This represents an increase of $7.4 million, or 9%, from the balance of the allowance for loan losses at December 31, 2007. During that period, our non-performing loans
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 11
increased from $68.1 million at December 31, 2007, to $164.8 million at September 30, 2008. The increase in non-performing loans resulted in some deterioration in certain of our asset quality ratios. However, there is not a direct relationship between the percentage changes in the various asset quality ratios and the percentage changes in the allowance for loan losses and resulting provision for loan losses recognized in any given period. The decline in our asset quality ratios is not indicative of a deficiency in our allowance for loan losses, since the calculation of the allowance considers the overall composition of the portfolio and levels of delinquencies. The increases in non-performing loans during any period are taken into account when determining the allowance for loan losses because the coverage ratios we apply to our non-performing loans are higher than the coverage ratios applied to our performing loans. Additionally, as required by our primary regulator, the Office of Thrift Supervision, we update our collateral values on one-to-four family loans which are 180 days past due. If the estimated fair value of the loan collateral less estimated selling costs is less than the recorded investment in the loan, a charge-off of the difference is recorded to reduce the loan to its fair value less estimated selling costs. Therefore certain losses inherent in our non-performing one-to-four family loans are being recognized at 180 days of delinquency and accordingly are charged off. There are no material assumptions relied on by management which may not be apparent in our disclosures or reflected in our asset quality ratios and activity in the allowance for loan losses.
Prior to the fourth quarter of 2007, our charge-offs had been very low and were primarily attributable to a small number of loans. The significant percentage increase in charge-offs during the six months ended June 30, 2008 represents a $4.5 million increase in charge-offs, compared to the year ended December 31, 2007, which represents 5% of the allowance for loan losses at June 30, 2008. The provision for loan losses recorded during the six months ended June 30, 2008 compared to the year ended December 31, 2007 increased $8.5 million, which represents 10% of the allowance for loan losses at June 30, 2008. The change in the allowance for loan losses reflects the combined impact of charge-offs and the provision for loan losses recorded.
We will, in future filings, revise our discussion in “Results of Operations – Provision for Loan Losses” to include the text in the second half of the third paragraph of this response beginning with the following sentence: “The increase in non-performing loans resulted in some deterioration in certain of our asset quality ratios.”
Asset Quality, page 67
10. | Please revise the last paragraph on page 68 to quantify the amount of loans outstanding as of the balance sheet dates presented that had FICO scores of 660 or below. |
As we are a portfolio lender, we underwrite our loans considering all credit criteria, as well as collateral value, and do not base our underwriting decisions solely on FICO scores. Although FICO scores are considered as part of our underwriting process, they have not always been recorded on our mortgage loan system and are not available for all of the one-to-four family loans on our mortgage loan system. However, substantially all of our one-to-four family loans originated since March 2005 have credit scores available on our mortgage loan system. One-to-four family loans, with credit scores available on our mortgage loan system, which had FICO scores of 660 and below at the date
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 12
of origination totaled $716.0 million, or 8% of one-to-four family mortgage loans with credit scores available on our mortgage loan system, at December 31, 2007 and $765.8 million, or 11%, at December 31, 2006. At both dates, such loans had a weighted average loan-to-value ratio, based on current principal balance and original appraised value, of 70%. We believe the aforementioned loans, when originated, were amply collateralized and otherwise conformed to our prime lending standards and do not present a greater risk of collectibility or other asset quality risk compared to loans to other borrowers with higher credit scores.
Since we do not consider the loans in our portfolio with FICO scores of 660 or below to be subprime loans, we believe that quantifying those loans in our disclosure would not provide meaningful and decision-useful information to the reader. Additionally, we do not believe the total amount of such loans is material or that it represents a significant risk concentration for disclosure purposes pursuant to AICPA SOP 94-6. As such, we do not believe we need to revise the December 31, 2007 Form 10-K.
Allowance for Loan Losses, page 72
11. | Please tell us and in future filings revise your discussions regarding the table at the bottom of page 72 to more clearly explain why you reduced the amounts allocated to Consumer and other loans from $13.8 million at December 31, 2005 to $11.4 million at December 31, 2006, with a further reduction to $7.5 million at December 31, 2007. |
The decreases in our allowance for loan losses allocated to consumer and other loans relate primarily to the decreases in this portfolio, both in balance and as a percentage of the total loan portfolio as well as the low level of actual charge-offs. Consumer and other loans totaled $356.8 million, or 2% of the total loan portfolio, at December 31, 2007; $430.8 million, or 3% of the total loan portfolio, at December 31, 2006; and $502.5 million, or 4% of the total loan portfolio, at December 31, 2005. Home equity lines of credit represented approximately 90% of total consumer and other loans at December 31, 2007, 2006 and 2005. Additionally, consumer and other loan charge-offs and non-performing loans have been relatively stable and low over the past few years relative to the size of this portfolio. Net charge-offs on consumer and other loan totaled $588,000, representing 15 basis points of average consumer and other loans outstanding, for the year ended December 31, 2007 and $374,000, representing 7 basis points of average consumer and other loans outstanding, for the year ended December 31, 2005. For the year ended December 31, 2006, we had net recoveries on consumer and other loans totaling $6,000. Non-performing consumer and other loans totaled $1.5 million, or 0.4% of consumer and other loans outstanding, at December 31, 2007; $818,000, or 0.2% of consumer and other loans outstanding, at December 31, 2006; and $500,000, or 0.1% of consumer and other loans outstanding, at December 31, 2005.
We will, in future filings, enhance our discussion regarding the table which details the allocation of the allowance for loan losses by loan category to address material changes in the allocation by loan category from year to year.
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 13
Form 10-Q for the Period Ended March 31, 2008
Financial Statements
General
12. | Due to the material nature of these disclosures to your business and trends experienced, please revise future interim filings to include footnote disclosure updating the securities footnote, the loans receivable footnote and the allowance for loan losses footnote. |
We incorporate updates to the information in our securities footnote, loans receivable footnote and allowance for loan losses footnote as reported in our Annual Report on Form 10-K throughout the MD&A in our Quarterly Reports on Form 10-Q. These updates include, but are not limited to, loan portfolio composition tables as of the balance sheet dates, a table of the changes in our allowance for loan losses from the prior year end, a securities portfolio table as of the balance sheet dates and information regarding gross unrealized losses in our securities portfolio.
We will, in future interim filings, move the loan portfolio and securities portfolio tables on pages 24 and 25 of the March 31, 2008 Quarterly Report on Form 10-Q into the footnotes and exclude them from the MD&A. We will also add a table of the changes in our allowance for loan losses, in summary form, into the footnotes and maintain the detailed table of changes in our allowance for loan losses on page 40 of the March 31, 2008 Quarterly Report on Form 10-Q in the Asset Quality section of the MD&A. Additionally, we will include a discussion, as deemed necessary, of other material updates from our year end disclosures related to these items.
13. | It appears you are presenting capitalized mortgage servicing rights in the operating section of the statements of cash flows. Please tell us how you acquire these rights, and tell us the accounting literature you relied on for this presentation. Generally, mortgage servicing rights arise when loans are sold and the rights are separated from the loans, resulting in a non-cash event. |
We acquire our mortgage servicing rights (MSR) through the sale of loans with servicing retained. The presentation of capitalized MSR within the same line item as MSR amortization and valuation allowance adjustments resulted in a minor overstatement of proceeds from sales of loans held-for-sale and an offsetting understatement of MSR amortization and valuation allowance adjustments, by the capitalized MSR amount. We based this presentation on the immateriality of the capitalized MSR and the fact that there was no effect on the amount reported as net cash provided by operating activities. Capitalized MSR totaled $196,000 for the three months ended March 31, 2008 and $1.4 million for the year ended December 31, 2007. MSR, net, totaled $12.3 million at March 31, 2008 and $12.9 million at December 31, 2007.
We will, in future filings, reflect capitalized MSR as an adjustment in determining the amount reported as proceeds from sales of loans held-for-sale rather than including the capitalized MSR in the same line item as MSR amortization and valuation allowance adjustments.
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 14
Note 6. Fair Value Measurements of Assets and Liabilities, page 8
14. | We note your disclosure that you obtain pricing for your mortgage backed securities, which comprise approximately 94% of your available-for-sale securities, from nationally recognized pricing services and third party brokers who provide prices which are categorized as Level 2 as quoted prices in active markets for identical assets are generally not available for the majority of securities in your portfolio. Paragraph 22 of SFAS 157 indicates that the level in the fair value hierarchy within which the fair value measurement in its entirety falls is determined based on the lowest level input that is significant to the fair value measurement in its entirety, and thus it is unrelated to whether a nationally recognized pricing service or a third party is used to price the financial instrument. Thus, classification is based upon the assumptions and inputs used to value the instrument and whether they are based on market observable inputs, or unobservable inputs. Please address the following: |
| · | Please tell us how you concluded that there was not a significant input into the valuation methodology used by pricing services and third party brokers that was based on unobservable data. |
| · | Please tell us and revise future filings to disclose the specific procedures management uses to validate the pricing received from pricing services and third party brokers, including the material assumptions used to arrive at the prices. |
| · | Please revise your future filings to ensure that your disclosure does not imply that the classification in the fair value hierarchy is based upon whether the value is determined based on a pricing service or a third party broker. Instead, please provide disclosure discussing the techniques used, the data used for the inputs/assumptions, and whether the inputs/assumptions are based on unobservable inputs, and how this information drove the classification in the fair value hierarchy. |
Management has reviewed the documentation provided to us by our independent pricing service regarding their analysis of SFAS No. 157, as well as their summary of inputs utilized by asset class and evaluation methodology summaries. Additionally, our pricing service has indicated, in writing, that if they do not have sufficient objectively verifiable information to continue to support a security’s valuation, they will discontinue evaluating the security until such information can be obtained. Our pricing service uses various modeling techniques to determine pricing for our mortgage-backed securities, including option pricing and discounted cash flow models. The inputs to these models include benchmark yields, reported trades, broker/dealer quotes, issuer spreads, benchmark securities, available trade information, bids, offers, reference data, monthly payment information and collateral performance. At September 30, 2008, 98% of our available-for-sale securities portfolio was comprised of GSE securities for which an active market exists for similar securities, making observable inputs readily available. Our Treasury Department analyzes changes in the pricing service fair values from month to month taking into consideration changes in market conditions including changes in mortgage spreads, changes in treasury yields and changes in generic
Mr. Kevin W. Vaughn
United States Securities and Exchange Commission
Page 15
pricing on 15 year and 30 year securities. Each month our Treasury Department conducts its own review of the estimated values of our fixed rate REMICs and CMOs available-for-sale which represent substantially all of these securities priced by our pricing service. The Treasury Department generates prices based upon a “spread matrix” approach for estimating values. Market spreads are obtained from independent third party firms who actively trade these types of securities. Any notable differences between the pricing service prices and “spread matrix” prices on individual securities are analyzed further by our Treasury Department. This additional analysis includes a review of prices provided by other independent parties, a yield analysis and review of average life changes using Bloomberg analytics and a review of historical pricing on the particular security. We have concluded, based upon our review of the information and prices provided by our pricing service, that the fair values of securities incorporate observable market inputs commonly used by buyers and sellers of these types of securities at the measurement date in orderly transactions between market participants, and, as such, represent Level 2 pricing within SFAS No. 157.
We will, in future filings, revise our disclosures to clearly indicate that the classification of our Level 2 fair values in the fair value hierarchy of SFAS No. 157 is not determined based solely on the fact that a pricing service was used to obtain those values. Additionally, we will add a discussion of the procedures management uses to validate the pricing received from our pricing service.
The Company acknowledges the following:
| ▪ | the Company is responsible for the adequacy and accuracy of the disclosure in the filing; |
| ▪ | staff comments or changes to disclosure in response to staff comments do not foreclose the Commission from taking any action with respect to the filing; and |
| ▪ | the Company may not assert staff comments as a defense in any proceeding initiated by the Commission or any person under the federal securities laws of the United States. |
Please contact the undersigned at (516) 327-7892 or (516) 327-7820, respectively, if you have any questions.
Sincerely,
| s/ Monte N. Redman | s/ Frank E. Fusco | |
| Monte N. Redman | Frank E. Fusco | |
| President and Chief Operating Officer | Executive Vice President, Treasurer | |
| Astoria Financial Corporation | and Chief Financial Officer | |
| Astoria Financial Corporation |