February 2, 2012
Via Email and EDGAR
United States Securities and Exchange Commission
Division of Corporation Finance
100 F Street N.E.
Washington, D.C. 20549
Attn: Ms. Suzanne Hayes, Assistant Director
Financial Services II
Room No. 3514
Telephone No. (202) 551-3675
hayess@sec.gov
Re: Federal Home Loan Bank of Boston (the “Bank”)
Response to December 27, 2011 Comment Letter
Form 10-K for the Fiscal Year Ended December 31, 2010
Filed March 18, 2011
File No. 000-51402
Dear Ms. Hayes:
We deliver this letter to you in response to the comment letter, dated December 27, 2011 (the “Comment Letter”), from the staff of the Division of Corporation Finance (the “Staff”) of the Securities and Exchange Commission (the “SEC”), on the Federal Home Loan Bank of Boston's (the “Bank's”) Annual Report on Form 10-K for the fiscal year ended December 31, 2010 (the “2010 Form 10-K”) filed with the SEC on March 18, 2011. Our response to the Comment Letter, along with a reproduction of each of the Staff's comments, follows. All page references refer to the 2010 Form 10-K. We appreciate the Staff's suggestions to further improve the Bank's disclosure filings.
Item 1. Business, page 3
General, page 3
Comment 1:
Please expand your disclosure in future filings to describe the Resolution Funding Corporation and Financing Corporation programs.
Response 1:
Overview of the Resolution Funding Corporation
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 2 of 19
As discussed on page 142, the Resolution Funding Corporation (“REFCorp”) is a corporation established by Congress in 1989 to provide funding for the resolution and disposition of certain insolvent savings and loans institutions. Officers, employees, and agents of the Office of Finance, an agent of the 12 Federal Home Loan Banks (the “FHLBanks”), are authorized to carry out the functions of REFCorp.
Material Impact of REFCorp on the Bank
As discussed beginning on page 23, the Bank (like the other FHLBanks) was obligated to make payments to REFCorp in the amount of 20 percent of net earnings after its Affordable Housing Program (“AHP”) expenses until August 5, 2011. The REFCorp contribution requirement was established by Congress in 1989 to provide funds to pay a portion of the interest on debt issued by the Resolution Trust Corporation that was used to assist failed savings and loan institutions. These interest payments totaled $300 million per year, or $75 million per quarter for the 12 FHLBanks through 1999. In 1999, the Gramm-Leach-Bliley Act of 1999 changed the annual assessment to a flat rate of 20 percent of net earnings (defined as net income determined in accordance with accounting principles generally accepted in the United States of America (“GAAP”) before the Bank's REFCorp assessment and after the Bank's AHP expense). Since 2000, the FHLBanks were required to make payments to REFCorp until the total amount of payments made was equivalent to a $300 million annual annuity with a final maturity date of April 15, 2030, which was the expiration date of the REFCorp obligation established in 1989. However, since the FHLBanks contributed a fixed percentage of their net earnings to REFCorp, the aggregate amounts paid frequently exceeded the required $75 million per quarter. As a result, on August 5, 2011, the Federal Housing Finance Agency, (the “FHFA”), the Bank's principal regulator, certified that the FHLBanks had fully satisfied their obligation to REFCorp with their payment based on their results as of June 30, 2011, as was first reported in Item 7.01 (Regulation FD) and related exhibits of the Bank's Current Report on Form 8-K filed with the SEC on August 5, 2011. Accordingly, the REFCorp contribution requirement has been fully discharged.
The Bank will include disclosure on these points in its Annual Report on Form 10-K for the fiscal year ended December 31, 2011 (the “2011 Form 10-K”), as set forth above and include similar disclosure in other, future Form 10-K filings as well as in other reports to the extent helpful, material, or otherwise required under applicable law and regulations, including the SEC's regulations.
The Financing Corporation
The Financing Corporation is a corporation established by Congress in 1987 originally intended to finance a rebuilding of the Federal Savings and Loan Insurance Corporation (the “FSLIC”). The FSLIC was an institution that administered deposit insurance for savings and loans institutions but became insolvent in the late 1980s and was abolished. The Financing Corporation subsequently served as a financing vehicle for the FSLIC Resolution Fund, which assumed the assets and liabilities of the FSLIC following its insolvency. The Financing Corporation originally was authorized to issue debt obligations to fund the FSLIC Resolution Fund, however, Congress terminated this authority in the early 1990s. Since that time, the Financing Corporation's sole purpose has been to manage the outstanding debt obligations that it issued. The functions of the Financing Corporation are unlikely to have a material impact on the Bank because the Bank has no continuing responsibilities to it other than, together with the other FHLBanks, to direct the Office of Finance to manage its activities.
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 3 of 19
Given the Financing Corporation's immateriality to the Bank, the Bank will delete references to it in its future filings.
AHP Assessment, page 24
Comment 2:
We note that in 2010 your Affordable Home Loan Program assessment of $11.8 million exceeded ten percent of your net income of $106.6 million. Please clarify whether this excess assessment was due to a shortfall from the assessments paid by other Federal Home Loan Banks or otherwise. To the extent that your assessment deviates from the ten percent formula in the future, please revise your future filings to explain the reason for the discrepancy.
Response 2:
For the year ending December 31, 2010, the AHP assessment was equal to 10 percent of net income, excluding interest expense on mandatorily redeemable stock, before the AHP assessment, but after the REFCorp assessment. As disclosed on page 24, and also in Note 15 - Affordable Housing Program, on page 172, the exclusion of interest expense associated with mandatorily redeemable capital stock is based on an advisory bulletin issued by the Federal Housing Finance Board, which was the regulator of the FHLBanks until July 30, 2008, upon the enactment of the Housing and Economic Recovery Act of 2008. This advisory bulletin remains in effect under the FHFA. Interest expense on mandatorily redeemable capital stock is equivalent to dividends paid on non-redeemable stock. However, for the year ending December 31, 2010, the Bank declared no dividends and interest expense on mandatorily redeemable capital stock was $0.
As such, the AHP assessment calculation for 2010 was determined as follows (dollars in thousands):
| Net Income Before Assessments | $ | 145,074 | ||
| Add back: Interest Expense on Mandatorily Redeemable Capital Stock | — | |||
| Less: REFCorp Assessment | (26,646 | ) | ||
| Net Income Basis for AHP Assessment | $ | 118,428 | ||
| 10% of Net Income Basis for AHP Assessment | $ | 11,843 | ||
Additionally, we have disclosed on pages 24 and 172, that there were no shortfalls in the required minimum $100 million annual AHP assessments by the 12 FHLBanks in 2010, 2009, and 2008. For the year ended December 31, 2010, the 12 FHLBanks accrued collective AHP assessments of $229 million.
Please note that the FHLBanks' REFCorp assessment obligation was concluded with the assessment for the quarter ending June 30, 2011, as discussed above in the response to Comment 1. Accordingly, for the full year ending December 31, 2011, the Bank's calculation of its AHP assessment will incorporate the accrual of REFCorp assessments through June 30, 2011. For years ending after December 31, 2011, the AHP assessment will be based simply on 10 percent of net income before assessments excluding interest expense on mandatorily redeemable capital stock.
If the AHP assessment should ever deviate from the foregoing explanation, the Bank will explain the discrepancy in the relevant filing(s).
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 4 of 19
Item 1A. Risk Factors, page 25
The loss of significant members could result in lower demand…, page 28
Comment 3:
Please expand your risk factor to identify the names of any members that account for more than ten percent of your advances.
Response 3:
We will expand the risk factor in future filings beginning with the 2011 Form 10-K to identify the names of members, if there are any, accounting for more than 10 percent of advances by par value to total par value of all Bank advances outstanding at the end of the Bank's fiscal year. At December 31, 2010, RBS Citizens, N.A. was the only member accounting for more than 10 percent of the Bank's outstanding advances by par value to total par value of all Bank advances outstanding, as set forth in the table captioned “Top Five Advance-Borrowing Members” on page 59.
Item 7. Management's Discussion and Analysis of Financial Condition and Results…, page 38
Executive Summary, page 39
Decline in Advances Balances, page 40
Comment 4:
We note your disclosure about the declining trend in the balance of advances to members. We also note that in March of 2011 FHFA Acting Director Edward DeMarco warned that large investments portfolios (as compared to advances) intended to generate added earnings is inconsistent with the purposes of the FHLBank System. We also note that in recent quarterly reports you have disclosed that management intends to maintain a level of investments at a level approximately equal to or less than 50 percent of total assets. Please expand your disclosure in the MD&A to address what changes you are making to your investments and to discuss the effect such changes would likely have on future earnings.
Response 4:
We are sensitive to the concerns raised by FHFA Acting Director DeMarco and note that our total investments have been less than 50 percent of our total assets for the last 10 years. However, we do not believe that our own investment guideline of maintaining total investments at a level equal to or less than 50 percent of total assets has had or is likely to have a material impact on our ability to carry out our mission or meet operating challenges that could arise due to market events because we can satisfy it, and expect to be able to continue to be able to satisfy it, by (if needed) divesting our short-term (and low-yielding) investments.
We maintain a portfolio of short-term, highly liquid, but very low-yielding investments predominantly based on i) statutory and regulatory liquidity requirements (detailed on page 95) and ii) potential member borrowing needs, as discussed on page 7. However, we are able to structure our finances to reduce our reliance on short-term investments to satisfy these needs and therefore can divest these low-yielding investments to satisfy our self-imposed 50 percent of total assets investment limitation. Because these
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 5 of 19
investments do not contribute meaningfully to our net income, their divestiture has not had, and is unlikely to have, a material impact on our net income. We have noted the divestiture of these investments in our quarterly reports on Form 10-Q throughout 2011 in connection with our 50 percent of total assets investment limitation.
Our long-term investments can make a more important contribution to our net income than our short-term investments. Although we have made changes to our portfolio of long-term investments in 2011, these changes have been driven primarily by other factors, most importantly our risk-return assessments of the investments. However, our self-imposed 50 percent of total assets investment limitation has not been a significant factor, and we do not expect it to be a significant factor in the future.
We will enhance disclosures in future filings starting with the 2011 Form 10-K to clarify that our self-imposed investment limitation has not had, and is unlikely to have, a material impact on our results of operations or financial condition. We will continue to disclose factors that have had and could reasonably be expected to have a material impact on our results of operations or financial condition, including any such factors arising from our investments.
Financial Condition, page 57
Mortgage Loans, page 78
Allowance for Credit Losses on Mortgage Loans, page 81
Comment 5:
We note your disclosure that during 2010 you changed key variables that drive the calculation of your allowance for loan losses, including the loss severity estimates applied to defaulted loans. Please confirm whether this was the primary driver of the significant provision recorded in 2010. Please also describe for us your process for updating the inputs to your allowance for loan loss model as it would appear that significant housing price deterioration also existed in periods prior to 2010, yet provisions of a similar magnitude were not recorded in those periods. Finally, please clarify for us your statement that the magnitude of the increase in 2010 reflects the fact that the increased loss projections would exceed the recoveries from any future credit enhancement fees you may withhold to offset such losses.
Response 5:
Increase in the Bank's Loan Loss Reserve At December 31, 2010
At December 31, 2010, the allowance for loan losses on our portfolio of Mortgage Partnership Finance® (“MPF”®) conventional loans (described below) was $8.7 million, representing a reserve equal to 0.29 percent of the MPF conventional loan portfolio. At December 31, 2009, the allowance for loan losses was $2.1 million, representing a reserve equal to 0.07 percent of the MPF conventional loan portfolio.
® “Mortgage Partnership Finance” and “MPF” are registered trademarks of the FHLBank of Chicago.
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 6 of 19
The primary driver of the increased loan loss reserve at December 31, 2010 over the loan loss reserve determined in prior periods was the use of increased loss severity estimates that we apply in our loan loss reserve estimation process, consistent with our disclosure on page 81. This factor accounted for approximately 70 percent of the increase. We used increased loss severity estimates based on the weakening performance of the Bank's portfolio of MPF conventional loans and continuing weakness in the U.S. housing economy. We increased the estimated loss severity upon default based on the deterioration in house prices and the existence of significant excess housing inventory in the U.S. While housing prices had begun to deteriorate in prior periods and excess housing inventory in the U.S. had begun to exist in prior periods, our loan loss reserves in prior periods were lower because of the differences between the performance of our portfolio of MPF conventional loans and other, comparable mortgage portfolios, as discussed below under - Our Process for Updating Inputs to Our Allowance for Loan Losses Model.
Overview of the Bank's Method for Determining the Allowance for Loan Losses
Our method for determining the allowance for loan loss reserves involves management judgment and is one of our critical accounting estimates, as described on page 109. The allowance is based entirely on our investments in MPF loans since these are the only whole loan investments that we make in accordance with FHFA regulations. Our method for determining the allowance is described beginning on page 159. For purposes of determining the allowance, we bifurcate our investments in MPF loans between a) MPF government loans that are either insured or guaranteed by the government (“MPF government loans”); and b) MPF conventional loans (“MPF conventional loans”), which are all of the Bank's other investments in MPF loans. Our MPF government loan portfolio totaled $322.8 million at December 31, 2010 and $335.9 million at December 31, 2009, and our MPF conventional loan portfolio totaled $2.9 billion at December 31, 2010, and $3.2 billion at December 31, 2009.
MPF government loans are either insured or guaranteed by U.S. agencies, as described on page 159. Based on the U.S. federal government agencies' insurance or guarantees, the Bank does not recognize any allowance for losses on such MPF loans.
Our method for determining the allowance on MPF conventional loans, described on page 160 in greater detail, considers various data observations such as past and current performance, loan portfolio characteristics, collateral-related characteristics, industry data, and prevailing economic conditions. The allowance also factors in available credit enhancement on MPF conventional loans, which can include primary mortgage insurance, supplemental mortgage insurance, member-provided credit enhancements (described below), and any other credit enhancement. The credit enhancement represents the participating financial institution's (the “PFI's”) continuing obligation to absorb a portion of the credit losses arising from PFI's master commitment(s), as described in greater detail on pages 14-15 and 159-160. A master commitment is a document which provides the general terms under which a PFI will deliver mortgage loans to us, including a maximum loan delivery amount, maximum credit enhancement, if applicable, and expiration date. We pay a credit-enhancement fee as described on page 15, and the fee can be withheld in certain instances, also described on page 15. Any withheld credit-enhancement fees can mitigate losses from our investments in MPF loans, as described on page 160. Accordingly, we also consider estimates of the amount of credit-enhancement fees that we expect to withhold in determining the allowance.
Additional information on our method for determining the allowance for loan losses, including the
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 7 of 19
collective and individual evaluation of our investments in MPF conventional loans, is set forth on page 160.
Our Process for Updating Inputs to Our Allowance for Loan Losses Method
We adjust the key inputs to our loan losses model so that projections of losses are properly aligned with market conditions as relevant to our investments in MPF conventional loans. As described on page 160, the key inputs to the model, in each case based on our investments in MPF conventional loans, include past and current performance (including portfolio delinquency and default migration statistics), overall portfolio performance characteristics (including loss mitigation features, especially credit enhancement, as discussed below under- Impact of Loss Mitigating Features on the Allowance for Loan Losses), collateral-related characteristics (such as property type and location), industry data (such as foreclosure timelines and associated costs), and prevailing economic conditions (including housing prices, the unemployment rate and mortgage loan interest rates).
Quarterly Updates
We update the following inputs in the loan loss reserve calculation at least once per quarter: current outstanding loan amounts, delinquency statistics of the portfolio, and related loss mitigation features (including credit-enhancement and estimated credit-enhancement fee-recovery amounts) for all master commitments. Additionally, we review the model's default migration factor on a quarterly basis to determine if the portfolio has exhibited any change in loans migrating from current to delinquent, or from delinquent to default and make adjustments as appropriate.
Annual Updates
Prior to 2011, we updated the loss severity estimates for all master commitments once per year, which was done during the fourth quarter of the year ended December 31, 2010. We determine these loss severity estimates using a third-party loan level loss projection model, which we validate every year and prior to implementation of a new version. We validate the model by reviewing the model's theoretical framework and data inputs and comparing the model's outputs against industry benchmarks. The model in effect as of December 31, 2009 (the “2009 Model”), was validated and confirmed acceptable for our use during the third quarter of 2009 and was later confirmed by a model validation performed by an independent third party. In 2010, we updated the model to the then-current model (the “2010 Model”) after validating and confirming that it was acceptable for our use. The 2010 Model had been substantially revised compared with the 2009 Model to better reflect deteriorating real estate market conditions. Particularly important in that revision was the use of extended foreclosure timelines and increased costs associated with foreclosures. The revised model also increased the weighting it applied to loan-to-value ratios.
Market conditions had begun deteriorating before 2010, and our key inputs driving loss estimates reflected this deterioration as early as 2008 and continuing through 2010. However, consistent with market data, the MPF conventional loans in which we had invested that were originated at the peak of housing prices (2006-2007) showed higher projected losses earlier than loans originated before 2006 as homeowner equity in these loans was eliminated the soonest after housing prices began to decline. For MPF conventional loans in which we had invested prior to 2006, the broad decline in house prices did not eliminate homeowner equity until more recently. Thus, projected losses on such loans originated prior to
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 8 of 19
2006 are, relatively, a more recent dynamic, and our loss projections capture this dynamic. These dynamics were appropriately reflected in the 2010 Model. As a result, the increase to our loan loss reserve during 2010 was overwhelmingly caused by higher loss severity estimates on such loans originated before 2006. Projected losses on MPF conventional loans in which we had invested from 2006 through 2008 had increased earlier, and continued to increase. The overall impact of higher per-loan loss severity projections affecting an increasing number of loans in the 2010 Model was an increase in projected loan losses that was reflected in an increase to our allowance for loan losses.
Starting with the quarter ended June 30, 2011, as part of our continuing effort to incorporate more precision into our estimation processes, we added a quarterly loss severity update for the 10 largest master commitments, reflecting a more current housing price index. These master commitments represent approximately 60 percent of our total master commitments by outstanding principal balance and therefore this update is likely to capture any important changes during interim reporting periods.
Impact of Loss Mitigating Features on the Allowance for Loan Losses
As noted above, withheld credit-enhancement fees can mitigate losses from our investments in MPF loans and therefore we consider our expectations by each master commitment for such withheld fees in determining the allowance for loan losses. More specifically we determine the amount of credit-enhancement fees available to mitigate losses as follows: 1) accrued credit-enhancement fees to be paid to PFIs; plus 2) projected credit-enhancement fees to be paid to the PFIs over the next 12 months; 3) minus any losses incurred or expected to be incurred.
The allowance for loan losses considers this factor in the context of the allocation of losses between us and the PFIs, which varies according to MPF conventional loan product (as noted above, there is no allowance for loan losses for MPF government loans) and is detailed on page 15. This loss allocation considers the first loss account and credit enhancement. See the discussion beginning on page 14 for a detailed discussion of the first loss account and loss priorities.
The following table demonstrates the impact on our estimate of the allowance for credit losses resulting from the loss mitigating features of MPF conventional loan products (dollars in thousands).
December 31, 2010 | December 31, 2009 | |||||||
| Estimate of credit losses | $ | 11,200 | $ | 3,993 | ||||
| Less: Losses in excess of the first loss account, to be absorbed by the PFI | (1,473 | ) | (348 | ) | ||||
| Less: Performance-based credit-enhancement fees available for recapture | (1,074 | ) | (1,545 | ) | ||||
| Allowance for credit losses | $ | 8,653 | $ | 2,100 | ||||
As shown in this table, the increase in the estimate of credit losses exceeded the increase in losses to be absorbed by the PFI and the increase in available performance-based credit-enhancement fees subject to recapture, as disclosed on page 81.
We will include disclosures on the foregoing points in the 2011 Form 10-K (including the immediately above table) and similar disclosures in other, future Form 10-K filings as well as in other reports to the extent helpful, material, or otherwise required under applicable law and regulations, including the SEC's regulations.
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 9 of 19
Item 8. Financial Statements
Note 20 - Fair Values, page 181
Comment 6:
We note your disclosure on page 187 related to significant inputs to your fair value measurements. You state that you obtain prices from up to four designated third-party pricing vendors when valuing your investment securities and that you test the values for reasonableness using specified tolerance thresholds. Please address the following:
Comment 6a:
Your disclosure states that you establish a price for each investment security using a formula that is based on the number of prices received. Please describe for us the formulas used in your fair value calculation (i.e., do you just take the average of the prices received?).
Response 6a:
In an effort to promote financial reporting consistency among the 12 FHLBanks, a MBS Pricing Governance Committee (the “Pricing Committee”) was formed as described on page 107. As a voting member, we provide input into the procedures considered by the Pricing Committee in determining fair values, and vote to adopt amendments to the procedures from time to time. We participate in the Pricing Committee as an important part of our continuing efforts to refine the valuation procedures for our investment securities and to enhance consistency among the FHLBanks on certain investment securities' valuation determinations. As we do each reporting period, for the period ended December 31, 2010, we reviewed the Pricing Committee's procedures and determined that they were reasonably designed to determine the prices that would have been received if the securities had been sold at the measurement date (“exit prices”). The Pricing Committee continues to review its processes to enhance its methodologies and incorporate best practices. As part of this effort, the Pricing Committee has adopted, and we will in turn employ, enhanced methodologies for the period ended December 31, 2011, as detailed in Response 7.
Vendor prices are used as part of a formula to calculate the fair value of our investment portfolios with the exception of our investments in housing-finance-agency (“HFA”) obligations. This formula, as of December 31, 2010, was as follows:
For each investment security that is subject to this methodology, we collected prices from each of the four designated and market-recognized pricing vendors to the extent available. We then established a “default price” for each investment security using a median calculation that was based upon the number of prices received:
| • | If four prices were received, the average of the middle two prices was used; |
| • | if three prices were received, the middle price was used; |
| • | if two prices were received, the average of the two prices was used; and |
| • | if one price was received, it was used subject to further assessment, as described below in Response 6b. |
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 10 of 19
At December 31, 2010, we received prices from all four pricing vendors on approximately 87 percent of our portfolio, and we received prices from at least three of the four pricing vendors on 96 percent of our portfolio. The fair values of our portfolios of investment securities for which we used vendor prices at December 31, 2010, include the following securities (dollars in thousands):
| Trading securities | ||||
| Short-term certificates of deposit | $ | 5,319,921 | ||
Government- and GSE-issued mortgage-backed securities (“Agency MBS”) | 259,820 | |||
| Total trading securities | $ | 5,579,741 | ||
| Available-for-sale securities | ||||
| Non-mortgage backed securities (“Non-MBS”) | $ | 5,332,130 | ||
| Agency MBS | 2,002,905 | |||
| Total available-for-sale securities | $ | 7,335,035 | ||
| Held-to-maturity securities | ||||
| Non-MBS | $ | 101,566 | ||
| Agency MBS | 4,405,444 | |||
| Private-label mortgage backed securities | 1,867,438 | |||
| Total held-to-maturity securities | $ | 6,374,448 | ||
Comment 6b:
More clearly describe the variance thresholds used to assess the reasonableness of the computed prices. For example, explain how significant discrepancies between prices received from your third-party pricing vendors are handled.
Response 6b:
The default price, as determined per the methodology described above in Response 6a, was used to calculate the security's fair value after testing it for reasonability by comparing the security's default price to the security's vendor prices. If the difference between the default price and the closest vendor price was more than that security's appropriate tolerance parameter, then that security's vendor prices were subject to further review.
As of December 31, 2010, the tolerance parameters were 7 points for private-label MBS and 2 points for our other investment securities that were valued using this methodology. The Pricing Committee recommended, and we in turn adopted, the tolerance parameters based on an FHLBank analysis that determined the range of typical observed price differences during historical time periods that were reflective of current market conditions. These tolerance parameters are a function of the observed dispersion among the prices received from the pricing vendors. The larger private-label MBS tolerance parameter reflects the increased market uncertainty over exit prices due to the continuing illiquidity and resulting limited price discovery for these securities, which is consistent with the wide range of reasonable collateral loss estimates and differing discount rates used by market participants and modelers that we have observed. The smaller tolerance parameters for the other securities that are subject to this valuation methodology reflect the greater certainty over exit prices due to these investments' liquidity and more
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 11 of 19
frequent opportunities for price discovery resulting in greater homogeneity in the range of prepayment-rate and yield-spread assumptions that we have observed.
Our methodology called for us to consider assigning a price other than the default price to a security (an “alternative price”) in cases i) when a default price was outside of the tolerance parameters, ii) other factors supported a price other than the default price even though the default price fell within the tolerance parameters, or iii) when only one vendor price or no vendor price was received. In such cases, our internal subject matter experts would develop and document their rationale for the selected price. Factors to be considered in determining whether or not to select an alternative price would include, but are not necessarily limited to:
| • | differences between the security's default price and prices of securities of similar characteristics in our portfolio, such as collateral type, credit quality, or expected weighted-average life or maturity; |
| • | a comparison of vendor price derived option-adjusted spread (“OAS”) for the security or projected yield to the market driven OAS or yields of similar securities in our portfolio; and |
| • | a comparison of the market assumption driven expected weighted-average life (“WAL”) for the security with the WALs of similar securities in our portfolio. |
For the period ended December 31, 2010, we did not have any security with a default price that was outside the pricing tolerance parameters, nor were any other factors deemed significant enough to warrant the use of an alternative price as the final price.
Our methodology provides that when we determine that an alternative price could be appropriate, we challenge the vendor(s) that provided us the prices that were used to determine the default price as a further step in our assessment. When a pricing vendor agrees that its price is incorrect, we exclude that price from our determination of that security's price. When a pricing vendor disagrees that its price is wrong, our internal subject matter experts determine whether or not to consider that price in the determination of that security's final price in the valuation process based on all information available to them, including the factors discussed above impacting the possible use of an alternative price.
To retain more market observations in the valuation of these securities, the Pricing Committee has adopted, and we are in turn employing, enhanced methodologies for the treatment of vendor prices that fall outside of our tolerance parameters for the period ended December 31, 2011. The new method employs the same median calculation approach as described in Response 6a, but it then subjects all vendor prices to a comparison versus the median (the “price variances”) to determine if any vendor price should be further reviewed and/or excluded from the default price calculation. The price variances versus the median are then compared to the security's applicable tolerance parameters. We believe that this methodology will result in the inclusion of more market observations in the valuation of our securities than our prior process that automatically eliminated the highest and lowest observations from the final price calculation when more than two vendor prices are received.
For the period ended December 31, 2011, the Pricing Committee has recommended, and we in turn have adopted, tolerance parameters of 10 points for private-label MBS and 3 points for Agency MBS. The tolerance parameters were increased from those used in the prior methodology, and were the result of an analysis that recognized the inherently wider price range that accompanies the use of a greater number of vendor prices in our pricing analysis. Further, we have adopted an internal set of tolerance parameters
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 12 of 19
ranging from 0.25 to 3 points, depending on the security type and term, for all other investments that are subject to this methodology. Any vendor price falling outside of these tolerance parameters will be subject to additional review and/or exclusion from the determination of the final price. The final price will generally be the average of the prices that fall within the tolerance parameters.
Comment 6c:
Tell us whether you obtain an independent auditor's report or other documentation from your pricing service providers regarding their controls over valuation of securities you classify in Level 2 of the fair value hierarchy to support the completeness and accuracy of the prices received.
Response 6c:
We obtain and review the information that our pricing vendors make available on the various ways in which they comply with market-accepted pricing techniques as well as their model validation and control processes. This information includes pricing methodology white papers and assumption overviews, and beginning in 2011, is supplemented by independent auditor's reports when available. In providing securities' prices, our pricing vendors generally use methods that consider, among other factors, benchmark yields, recent trades of similar securities, dealer indications, valuation models, benchmarking of like securities, sector groupings (based on collateral types, issues years, tranche types, etc.), and/or matrix pricing. In addition to obtaining an understanding of the valuation methods employed by the pricing vendors, our assessment of the completeness and accuracy of the prices received from the pricing vendors is based on our valuation methodologies described above in Responses 6a and 6b.
Comment 6d:
Explain to us whether any documentation from the pricing servicers or brokers contains caveats or disclaimers about the pricing information provided. If so, explain the procedures you performed to determine you have complied with GAAP relating to valuation and have effective ICFR.
Response 6d:
As a matter of course, our pricing vendors provide caveats and disclaimers about the pricing information they provide. Such caveats and disclaimers generally assure the user that the vendor believes the data is from sources it deems reliable, but that their accuracy and completeness cannot be guaranteed.
Given the subjective nature and the reasonable range that exists for many of the critical inputs to the pricing process, whether disclaimed or not, we believe it would be imprudent to rely solely on assertions or testimonials from the pricing vendors as to the accuracy of their price estimates. Instead, we further rely upon the screening, challenge, and exclusion controls that are in place that govern our valuation methodologies as discussed in Responses 6a through 6c and 6e.
Comment 6e:
Describe to us the frequency with which you challenge the pricing of any particular Level 2 financial instrument received from the service providers, the results of such challenges, and the impact the challenges had on your assessment of the service provider's internal controls.
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 13 of 19
Response 6e:
For the period ended December 31, 2010, we compared vendor prices against the tolerance parameters discussed in Response 6b at least once per quarter. During 2011, we compared vendor prices against the tolerance parameters at least once per month, and at each quarter-end. We ordinarily challenge our pricing vendors when a price falls outside of the tolerance parameters and/or if we determine a price should be excluded as result of our further analysis.
Instances when default prices fall out of the tolerance parameters have been infrequent as indicated by the results of our methodology for the period ended December 31, 2010, when we did not have any security with a default price that was outside the pricing tolerance parameters. However, when we determine that an alternative price could be appropriate, we challenge the vendor(s) that provided us the prices that were used to determine the default price as a further step in our assessment. When a pricing vendor agrees that its price is incorrect, we exclude that price from our determination of that security's price. When a pricing vendor disagrees that its price is wrong, our internal subject matter experts determine whether or not to consider that price in the determination of that security's final price in the valuation process based on all information available to them, including the factors discussed above impacting the possible use of an alternative price.
As part of our participation in the Pricing Committee, we annually review the performance of our pricing vendors including a review of their prices relative to the tolerance parameters and their reactions to our pricing challenges. We also review the available documentation and other information on our pricing vendors in assessing their controls, as discussed in Response 6c. We would neither select nor continue using the services of a pricing vendor where we had unresolved concerns about its prices or internal controls. Instances when default prices fall out of the tolerance parameters have been infrequent, which has given us a measure of additional comfort regarding our pricing vendors' internal controls. For purposes of our own controls, our process of challenging prices to test their reasonability enhances our comfort that the prices we establish for our securities are reasonably likely to be exit prices. Additionally, we supplement our controls over this process by testing prices even when they fall within the tolerance parameters when other factors support an alternative price, as discussed in Response 6b.
Comment 6f:
Tell us whether the third party pricing service providers provide you a list of the observable market data (including similar securities whose price they use or adjust) or model inputs that they use to price each security or class of securities in sufficient detail that you are able to assess whether the pricing methodology complies with ASC 820. If so, explain to us how you validate this data or otherwise confirm its completeness and accuracy.
Response 6f:
As outlined in Response 6c above, our pricing vendors publish and/or provide varying degrees of detail regarding their proprietary pricing processes. This detail includes, but is not limited to, lists of observable inputs obtained and/or imputed from trades of the same or similar securities as well as lists of modeled inputs for unobservable assumptions. For more complex securities these assumptions include credit,
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 14 of 19
collateral, and waterfall characteristics as well as OAS and voluntary prepayment estimates. Our pricing vendors do not typically provide detailed analysis or key data and assumptions on all securities they are asked to price, but they will provide data, rationale, and insight on prices that are challenged. Therefore, we believe the data that they gather combined with our own processes adequately supports our valuation of our securities that are subject to this process.
Comment 7:
Please also revise your future filings to provide the following additional disclosures:
| • | Describe the extent to which the pricing services are gathering observable market information as opposed to using unobservable inputs and/or proprietary models in making valuation judgments and determinations. To the extent proprietary models and observable inputs are used, discuss the internal controls you have in place to ensure the models and assumptions used by pricing services reflect those used by market participants. |
| • | Explain how you evaluate the accuracy and completeness of the observable inputs used by the pricing services in valuing securities in order to reach a conclusion that the fair value determination is consistent with ASC 820, and properly classified in the fair value hierarchy. |
| • | Describe the internal controls in place over the prices received from pricing services. |
Response 7:
We will include expanded disclosure in the 2011 Form 10-K as requested and include similar disclosure in future Form 10-K filings as well as other reports to the extent helpful, material, or otherwise required under applicable law and regulations, including the SEC's regulations. The substantive disclosures we will include in the 2011 Form 10-K are provided in the following indented text:
For the period ended December 31, 2011, we continued to employ the procedures to value our MBS that have been established by the Pricing Committee and also applied those procedures to value our non-MBS investments with the exception of floating-rate HFA obligations. We have reviewed the Pricing Committee's procedures and determined that they are reasonably designed to determine that the estimated prices were exit prices. Accordingly, the following descriptions of our procedures for the period ended December 31, 2011 are the same as those established by the Pricing Committee.
The four designated and market-recognized pricing vendors from whom we collected prices used various proprietary models. The inputs to those models were derived from various sources including, but not limited to: benchmark yields, reported trades, dealer indications, issuer spreads, benchmark securities, bids, offers, and other market-related data. Since many private-label MBS (and to a lesser extent other investment securities) do not trade on a daily basis, the pricing vendors used available information such as benchmark curves, benchmarking of like securities, sector groupings, and matrix pricing to determine the prices for individual securities. Each pricing vendor has an established challenge process in place for all securities valuations, which facilitates resolution of potentially erroneous prices as identified by our methodology.
In connection with the refinement of these procedures for the period ended December 31, 2011, we conducted reviews of the four pricing vendors to confirm and further augment our understanding of
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 15 of 19
the vendors' pricing processes, methodologies, and control procedures. To the extent available, we also reviewed the vendors' independent auditors' reports regarding the internal controls over their valuation processes.
In periods prior to the period ended December 31, 2011, we established a price (a “default price”) for each investment security using a formula that was based upon the number of prices received:
| • | If four prices were received, the average of the middle two prices was used; |
| • | if three prices were received, the middle price was used; |
| • | if two prices were received, the average of the two prices was used; and |
| • | if one price was received, it was used subject to further validation. |
We tested the default prices for reasonableness using specified tolerance parameters. We generally accepted the default prices that were within these parameters unless strong evidence suggested that using the formula-driven price would not be appropriate. Management conducted further analysis of securities' prices when i) preliminary estimated fair values were outside the tolerance parameters, ii) when prices were within the tolerance parameters but management observed anomalous price behavior, or iii) when only a single vendor price or no vendor price was received. Such further analysis included, but was not limited to:
| • | differences between the security's default price and prices of securities of similar characteristics in our portfolio, such as collateral type, credit quality, or expected weighted-average life or maturity; |
| • | a comparison of vendor price derived OAS or projected yield for the securities to the market-driven OAS or yields of similar securities in our portfolio; and |
| • | a comparison of the security's market assumption driven expected WAL to the WALs of similar securities in our portfolio. |
Beginning with the period ended December 31, 2011, the Bank refined its method for estimating the fair values of investment securities. The FHLBanks' revised valuation technique first requires the establishment of a “median price” for each security using a formula that is based upon the number of vendor prices received:
| • | If four prices are received, the middle two prices are averaged to establish a median price; |
| • | if three prices are received, the middle price is the median price; |
| • | if two prices are received, the prices are averaged to establish a median price; and |
| • | if one price is received, it is the median price (and also the final price), subject to further validation, consistent with the evaluation of “outliers” as discussed below. |
In the cases where two or more vendor prices are obtained, each vendor price is then compared to the median price defined above. Differences between all vendor prices received and the median price are then compared to the security's applicable tolerance parameters. The vendor prices that are within the security's tolerance parameters are averaged to arrive at the security's default price. In cases where one or more vendor prices have differences greater than the applicable tolerance parameter, these out-of-tolerance prices (the “outliers”) are subject to further analysis. Further analysis to exclude a vendor price or to determine if an outlier is a better estimate of fair value includes but is not limited to, comparison to i) prices provided by a fifth, third-party recognized valuation service, ii) observed
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 16 of 19
market prices for similar securities, and/or iii) dealer indications. Management makes this determination in consultation with the Bank's subject matter experts. In the case where management determines the outlier is more representative of the security's price, then the outlier price will be used as the final price rather than the default price. If, on the other hand, management determines that an outlier (or outliers) is (are) in fact not representative of fair value, then the outlier(s) will be excluded from the default price calculation, and the default price is then used as the final price. In all cases, the final price is used to determine the fair value of the security.
As an additional step, the Bank reviewed the final fair-value estimates of its private-label MBS as of December 31, 2011 for reasonableness using an implied yield test. The Bank calculated an implied yield for each of its private-label MBS using the estimated fair value derived from the process described above and the security's projected cash flows from the Banks' other-than-temporary-impairment assessment process and compared such yield to the market yield for comparable securities according to dealers and other third-party sources to the extent comparable market yield data was available. Management evaluated significant variances in conjunction with all of the other available pricing information to determine whether an adjustment to the fair-value estimate derived from the process described above was necessary.
As of December 31, 2011, four vendor prices were received for 90 percent of the Bank's investment securities and the final prices for substantially all of those securities were computed by averaging the four prices. Substantially all of the Bank's securities were priced by at least three vendors. The revised valuation technique did not have a significant impact on the estimated fair values of the Bank's investment securities as of December 31, 2011, compared with the prior valuation technique. Based on the Bank's reviews of the pricing methods and controls employed by the third-party pricing vendors and the relative lack of dispersion among the vendor prices (or, in those instances in which there were outliers or significant yield variances, the Bank's additional analyses), the Bank believes the final prices it used are reasonably likely to be exit prices and further that the fair-value measurements are classified appropriately in the fair-value hierarchy.
We will include the foregoing (or substantially similar) disclosure in the 2011 Form 10-K as requested and include similar disclosure in other, future Form 10-K filings as well as in other reports to the extent helpful, material, or otherwise required under applicable law and regulations, including the SEC's regulations.
As noted above, the revised valuation technique did not have a significant impact on the estimated fair values of our investment securities as of December 31, 2011. The differences between the prior valuation technique and the revised valuation technique on our investment securities valued using pricing vendor prices as of December 31, 2011, are provided in the following table (dollars in thousands):
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 17 of 19
| Fair Value using Prior Valuation Technique | Fair Value using Revised Valuation Technique | Difference | ||||||||||||
| Trading securities | ||||||||||||||
| Agency MBS | $ | 273,928 | $ | 274,164 | $ | 236 | 0.09 | % | ||||||
| Available-for-sale securities | ||||||||||||||
| Non-MBS | 4,100,067 | 4,100,819 | 752 | 0.02 | % | |||||||||
| Agency MBS | 1,179,689 | 1,179,380 | (309 | ) | (0.03 | )% | ||||||||
| Held-to-maturity securities | ||||||||||||||
| Non-MBS | 110,967 | 110,951 | (16 | ) | (0.01 | )% | ||||||||
| Agency MBS | 4,977,523 | 4,976,064 | (1,459 | ) | (0.03 | )% | ||||||||
| Private-label MBS | 1,425,219 | 1,425,084 | (135 | ) | (0.01 | )% | ||||||||
Item 10. Directors, Executive Officers and Corporate Governance, page 196
Member Director and Member Director Nominee Requirements and Nominations, page 197
Comment 8:
We note that you disclosed for each independent director the specific experience, qualifications, attributes or skills that led to the conclusion that the person should serve as your director, but did not provide similar disclosure for your member directors. Please provide for each member director the disclosure required by Item 401(e) of Regulation S-K, or tell us why you believe you are not required to do so. Please see our guidance in Regulation S-K Compliance and Disclosure Interpretation 116.09.
Response 8:
We believe that the disclosure required by Item 401(e) of Regulation S-K is not applicable to the Bank's member directors due to the unique nomination and election process dictated by federal law and by FHFA regulations for all FHLBanks. Unlike the director nomination and election process followed by most other corporate entities, the Bank has no control over the candidates who are nominated and elected as member directors, and, indeed, is prohibited by law from exerting any influence over that process. Nominations for each open seat may be made solely by member institutions located in the applicable state, and multiple candidates may be nominated for each open seat. The only eligibility requirements for a nominee to be added to the ballot for an open member directorship are that the nominee be a U.S. citizen, not be a member of the Bank's management, and be a director or officer of a member institution located in the applicable state. Moreover, the solicitation of proxies is not permitted; elections are conducted through a secret ballot process whereby the Bank has no way of knowing which candidate received the most votes until after the balloting deadline has passed. As a result, the Bank is not in a position to know what factors its member institutions considered in nominating candidates for member directorships or in voting to elect the Bank's member directors and therefore cannot disclose those factors.
We explain the member director nomination and election process and why we are unable to disclose the factors that were considered in nominating the candidates who were elected as member directors on pages
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 18 of 19
196 and 197.
We also acknowledge Regulation S-K Compliance and Disclosure Interpretation 116.09 but believe that this interpretation is also inapplicable. This interpretation provides that when a director is appointed by holders of a class of preferred stock, the company may either provide the same information about this director as it would for directors nominated by the board or disclose that the preferred shareholder has advised the company that the shareholder has appointed this director because of [the Item 401(e) information provided to the company by the shareholder that the company would then include in its filing]. Unlike preferred stockholders who have a right to appoint a director to represent a class of preferred stock, no stockholder of the Bank has any such appointive power. Any member institution/stockholder of the Bank as of the applicable record date may, if it wishes, nominate one of its eligible directors or officers to appear on the ballot for the member director election for the state in which it is located. However, having the ability to place a candidate on a ballot is not tantamount to having the power to appoint a director, as appearing on the ballot by no means guarantees a candidate's election as a member director. FHFA regulations do not require member institutions to communicate to the Bank the reasons for their nominations, and we have no power to require them to do so1. Moreover, even if we could obtain information as to why various candidates were nominated to run for election, we would have absolutely no way of knowing why the winning nominee (rather than any other nominee) received the plurality of votes from the voting members in a particular state.
Notwithstanding the foregoing, when the Bank's board of directors has a mid-term vacancy in a member directorship, FHFA regulations require the remaining directors to elect a director to fill the vacancy. In those instances, we can know what was considered in electing such directors, and have provided the disclosure required by Item 401(e) of Regulation S-K, as we do with independent directorships. An example of this is the disclosure provided on John F. Treanor, who was elected by the board on December 17, 2010 to fill the remainder of a vacant Rhode Island member directorship, as discussed on page 197. We will continue this practice in future filings to comply with Item 401(e) of Regulation S-K to the maximum extent possible.
_____________________________________________________________________________________
________________________
1 Companies that issue preferred stock typically do so pursuant to a stock purchase agreement whereby the company can contractually require a preferred stockholder with appointive powers to disclose the information that will enable the company to comply with Item 401(e) of Regulation S-K and Interpretation 116.09.
Ms. Suzanne Hayes, Assistant Director
United States Securities and Exchange Commission
February 2, 2012
Page 19 of 19
We acknowledge that:
| • | the Bank is responsible for the adequacy and accuracy of the disclosure in the filings; |
| • | Staff comments or changes to disclosure in response to Staff comments do not foreclose the Commission from taking any action with respect to the filings; and |
| • | the Bank 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. |
We have taken great efforts to respond to the Comment Letter in a timely and complete manner.
Please feel free to contact me at (617) 292-9624, Carol Hempfling Pratt, General Counsel, at (617) 292-9660, or Michael Arbogast, Corporate and Securities Counsel, at (617) 292-9776 should you wish to discuss these matters further.
Sincerely,
/s/ FRANK NITKIEWICZ |
| Frank Nitkiewicz |
| Executive Vice President and Chief Financial Officer |
cc: Angela Connell (connella@sec.gov)
Brittany Ebbertt (ebbertt@sec.gov)
Sebastian Gomez Abero (gomezaberos@sec.gov)
Celia Soehner (soehnerc@sec.gov)