Dale A. Thatcher EVP, Treasurer & Chief Financial Officer Selective Insurance Group, Inc. 40 Wantage Avenue Branchville, New Jersey 07890 Tel: (973) 948-1774 Fax: (973) 948-0282 E-mail: dale.thatcher@selective.com |
December 15, 2009
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
100 F Street, N.E.
Mail Stop 4720
Washington, D.C. 20549
RE: | Selective Insurance Group, Inc. |
Form 10-K for Fiscal Year Ended December 31, 2008
Form 10-Q for the quarter Ended June 30, 2009
Form 10-Q for the quarter Ended September 30, 2009
File No. 1-33067
Dear Mr. Rosenberg:
Thank you for your letter of December 1st. As requested, this letter is formatted so our responses correspond with your comments, which are in bold type. The capitalized terms we use in this letter have the meanings defined in our Annual Report on Form 10-K for the fiscal year ended December 31, 2008.
Form 10-K for the fiscal year ended December 31, 2008
Notes to ConsoIidated Financial Statements
Note 4 Investments, page 93
1. | Refer to your response to our comment three. With regards to your distressed debt strategy that includes a fund of funds component, it is our understanding that there are different types of “fund of funds,” each investing in a different type of collective investment scheme. Please clarify the types of fund of funds in which the company invests and the related investment scheme. |
SELECTIVE RESPONSE:
The following proposed prospective disclosure in our Form 10-K for the year ended December 31, 2009 will be added:
The fund of funds component of our distressed debt strategy, which makes up approximately $6.2 million of our distressed debt strategy, encompasses a number of strategies that generally fall into one of the following broad categories:
Distressed Debt Funds - Trading-Focused
These funds focus on buying and selling debt of distressed companies ("Distressed Debt").
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 2
Distressed Debt Funds - Restructuring-Focused
These funds focus on acquiring Distressed Debt with the intent of converting it into equity in a restructuring and taking control of the company.
Special Situations Funds
These funds pursue strategies that seek to take advantage of dislocations or opportunities in the market that are often related to, or are derivatives of, distressed investing. Special situations are often event-driven and characterized by complexity, market inefficiency, and excess risk premiums.
Private Equity Funds - Turnaround-Focused
These funds are a subset of private equity funds focused on investing in under-performing or distressed companies. These funds generally create value by acquiring the equity of these companies, in certain cases out of bankruptcy, and effecting operational turnarounds or financial restructuring.
Form 10-Q for the quarter ended June 30, 2009
Results of Operations & Related Information by Segment
Review of Underwriting Results by Line of Business, Page 36
2. | Refer to your response to our comment eight. You acknowledge that the guidance in FASB ASC 944-60-25-3 requires insurance contracts to be grouped consistent with the manner in which you acquire, service and measure profitability of your insurance contracts to determine if a premium deficiency exists. Please address the following: |
· | For each of your lines within the commercial and personal lines, in which you present net premiums earned in Note 9, please tell us how you acquire and service these contracts. Please explain why it is appropriate to group these contracts for the purpose of evaluating premium deficiencies. |
· | Although you state that you measure the profitability of your insurance contracts at your insurance operations segment level, you separately disclose, in MD&A, the profitability of your commercial and personal lines as well as for certain lines within commercial. For each of these lines please tell us how you use these profitability measures including to what extent you use it in pricing each line's contracts. In addition, provide us similar information for any other line within the commercial and personal lines for which you present net premiums earned in Note 9 and compute a profitability measure. |
SELECTIVE RESPONSE:
For purposes of determining how to group our insurance contracts in accordance with FASB ASC 944-60-25-3, we considered, and continue to consider, many of the same facts that we analyze when reviewing our operating segments under FASB ASC 280, “Segment Reporting.” We believe the underlying principle for both of these standards is consistent and requires us to determine how we manage the way we acquire, service, and analyze the profitability of our business.
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 3
The distribution of our products, the net premiums earned of which are outlined in Note 9 to the financial statements, centers around the independent agency distribution channel. Focusing on the broad insurance needs of our agency’s customers is extremely important, regardless of line of business. While we do price accounts on a granular level, this is largely a function of regulatory requirements and the competitive landscape within which we operate. Pricing at a granular level allows us to fulfill our regulatory reporting requirements and allows our customers to compare product offerings from different property and casualty insurance companies. In addition, pricing and profitability information at a granular level is necessary for the financial statement reader, as well as external financial analysts, to compare us with our competitors in the industry. However, our marketing efforts for all of our product lines revolve around independent agencies and their touch points with our shared customers, the policyholders. We service our agency distribution channel through our field model, which includes agency management specialists, loss control representatives, claim management specialists and our Underwriting and Claims Service Centers. None of these roles are delineated by line of business, rather, each function services the entire set of insurance contracts acquired through each agency. Furthermore, we measure profitability of our agents and calculate their compensation based on overall insurance results, not by line of business. All employees of the Company, including senior management, are incented based on overall insurance results, not by line of business. In our opinion, incentive compensation structure is one of the clearest and most objective methods of ultimately determining how a company manages the profitability of their business.
For these reasons, we believe that it is appropriate to group insurance contracts at our total Insurance Operations level when evaluating premium deficiency; despite our willingness to disclose additional information regarding our product lines and their corresponding profitability. We believe generally accepted accounting principles support the inclusion of these additional disclosures in our notes to the financial statements and in our Management Discussion and Analysis, despite the fact that that the information provided is at a more granular level than the basis on which we acquire these insurance contracts. Specifically, FASB ASC 280-10-05-5 provides that, “to provide some comparability between public entities, this Subtopic requires that an entity report certain information about the revenues that it derives from each of its products and services (or groups of similar products and services) … regardless of how the entity is organized. As a consequence, some entities are likely to be required to provide limited information that may not be used for making operating decisions and assessing performance.” Further, FASB ASC 280-10-45-2 provides that “nothing in this Subtopic is intended to discourage a public entity from reporting additional information specific to that entity or to a particular line of business that may contribute to an understanding of the entity.” We believe this additional disclosure is important to the financial statement reader, as well as external financial analysts, to understand and compare our results to others within our industry and we provided this additional information, in part, in response to previous Commission Staff requests.
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 4
Form 10-Q for the quarter ended September 30, 2009
Notes to Unaudited Interim Consolidated Financial Statement
Note 6, Investments. Page 9
3. | Please revise your disclosure provided in response to our comment five to include the weighted-average assumed conditional default rates in addition to the range you disclose. At a minimum, the ranges provided for your Alt-A fixed structured securities and your Alt-A hybrid structured securities appear to be sufficiently wide to limit their usefulness without an understanding of the weighted-average assumed conditional default rates. |
SELECTIVE RESPONSE:
To the extent that the disclosure on page 16 is applicable and will be presented in our Form 10-K for the year ended December 31, 2009, our proposed prospective disclosure is as follows:
For structured securities, we utilized underlying data for each security, including information from credit agencies, to determine projected future cash flows. These projections included base case and stress testing scenarios that modify expected default rates, loss severities, and prepayment assumptions based on type and vintage. We use conditional default rates up to 2.50 in the stress testing scenarios of our CMBS portfolio. Generally the range of the conditional default rate assumptions for our other structured securities is as follows:
· | Alternative-A securities (“Alt-A”) fixed structured securities: 0.50 – 6.00 |
· | Alt-A hybrid structured securities: 1.00 – 7.00 |
· | All other fixed structured securities: 0.07 – 1.00 |
· | All other hybrid structured securities: 0.33 – 1.50 |
Our Alt-A securities, which were comprised of 11 different securities, had an unrealized/unrecognized balance of approximately $7.9 million, the corresponding fair value of which represents less than one percent of the fair value of our fixed maturity security investment portfolio as of September 30, 2009. The weighted average conditional default rates for our Alt-A fixed structured securities and Alt-A hybrid structured securities applied during our stress testing as of September 30, 2009 was 3.00 and 3.95, respectively.
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 5
Form 10-Q for the quarter ended September 30, 2009
Notes to Unaudited Interim Consolidated Financial Statement
Note 6, Investments. Page 9
4. | On page 12, you provide disclosure as of September 30, 2009 consistent with the information provided in your response to our comment six as of June 30, 2009. In this disclosure, you indicate that you performed a test that stressed 14 HTM CMBS securities comprising a $30.7 million or 86% decline from carrying value under various scenarios with loss severities that generally ranged from 20% to 50%, based on loan-to-value ratios, as well as conditional default rates that generally ranged from 1.0 to 2.5. It remains unclear to us why the 86% average decline in value is not indicative of an other-than-temporary impairment considering that it appears that many of these securities have been in an unrealized loss position for more than 12 months. In addition, on page 53 you disclose the unrealized losses on the five individual security holdings with the largest unrealized balances ranging from 70% to 89%. Please explain to us how the large discounts from your cost bases on these securities are indicative of interest spreads in the market. In your response, please ensure that you address each of the following: |
· | For each of the 14 HTM CMBS securities you identify, please describe what "stressed these 14 securities" means and explain why the use of assumed loss severities ranging from 20% to 50%, based on loan-to-value ratios, is appropriate. Please also describe how the conditional default rates relate to loss severity. For example, describe how loss severities of 20% to 50% result from conditional default rates of 1.0 to 2.5. Please explain why assumed loss severities ranging from 20% to 50% do not appear to result in credit losses requiring impairment. |
· | Regarding your disclosure "generally these securities have experienced increased collateral support over origination," describe how this contributed to your conclusion that no credit impairment charge was required. Also describe how collateral support is considered in the determination of the fair value of the HTM CMBS securities. In this regard, please explain why the disclosed discounts up to 89% from amortized cost are not indicative of some amount of credit loss. |
· | You indicate that "generally these securities have loan-to-value ratios that support the valuation.'' Please revise to clarify what valuation you are referring to in "support the valuation." In this regard, are you referring to the fair value, which includes an average 86% discount, or to the principal value of the securities. In addition, please revise your disclosure to include a weighted average and range of loan-to-value ratios of your HTM CMBS. |
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 6
SELECTIVE RESPONSE:
To the extent that the discussion on page 12 is applicable and will be presented in our Form 10-K for the year ended December 31, 2009, our proposed prospective disclosure will be enhanced as follows in response to the third bullet of your comment:
“…Furthermore, we considered the following facts and circumstances: (i) these securities have experienced low delinquencies and in certain cases no losses to date; (ii) generally these securities have experienced increased collateral support over origination; and (iii) generally these securities have loan-to-value ratios that support the principal value of the securities. These loan-to-value ratios ranged from 37% to 83%, with a weighted average based on carry value of 66%.”
In addition, the following discussion, which is not part of the proposed prospective disclosure above, is provided in response to your specific questions regarding the 14 HTM CMBS securities deemed not to be other-than-temporarily impaired as of September 30, 2009. Please note that the five securities with the largest unrealized/unrecognized balances that are disclosed on page 53 of our September 30, 2009 Form 10-Q are included within the 14 HTM CMBS securities discussed on page 12 of that report.
There is currently a large market discount for subordinated CMBS, in addition to a continued market discount for highly structured and illiquid CMBS, especially for Re-REMICs, which continues to impact valuations. We believe these discounts are often irrespective of underlying credit characteristics of individual securities and instead represent a temporary imbalance between supply and demand brought on by the broader turmoil in the financial markets. In the case of these 14 securities, it is our belief that based on this imbalance, current valuations are not necessarily reflective of the underlying credit metrics, or the underlying ability of the collateral to repay. Collateral support is an important factor in evaluating all structured securities as it represents insulation from principal losses to the security. However, the statement in our disclosure about collateral support generally increasing was referring to all structured securities and it applies more to RMBS than CMBS. Collateral support generally increases as a percentage of total collateral on structured RMBS securities due to principal pay downs. Collateral support is a significant factor in evaluating the present value of future cash flows of these 14 CMBSs; however, CMBS deals do not generally require principal payments in the early years (i.e., the majority of our CMBS do not start to pay principal until year ten) so we do not see a significant increase in the credit support of these 14 CMBS. We are not currently seeing significant deterioration of credit support on these CMBS to date either. Some of our CMBS do require balloon payments, resulting in some refinancing risk, which we have considered in our evaluation of future cash flows. As such, valuations alone cannot be used to assess potential impairment. Using the process disclosed in our Form 10-Q for the quarter ended September 30, 2009 and as further described below, we concluded that the present value of estimated future cash flows was not less than amortized cost for these 14 securities.
Of these 14 securities, six securities comprised $3.1 million of the total $30.7 million unrealized/unrecognized loss position at September 30, 2009. These six securities had an average loss position of only 14% from the amortized cost of these securities; therefore we will not specifically address these six securities herein.
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 7
The drivers behind the 86% decline in the 14 HTM CMBS securities were eight securities that are listed in the table below. These eight securities comprised the remaining $27.6 million unrealized/unrecognized loss position at quarter end. Further discussions regarding both our general approach to evaluating these eight securities for potential credit losses, as well as specific discussions for each of these eight securities, are provided following the table.
As of September 30, 2009 | 12 months or longer | |||
($ in thousands) | Fair Value | Total Unrealized Unrecognized Losses | Loan Loss Severity | “First Loss” Conditional Default Rate |
GS Mortgage Securities | $ 136 | (728) | 44.2 | 1.95 |
ACT Depositor Corp | 253 | (2,562) | 50.0 | 4.54 |
GS Mortgage Securities Corp II* | 2,528 | (7,092) | 50.0 | 3.09 |
JP Morgan Chase Comm Mtg Sec* | 500 | (3,392) | 50.0 | 2.10 |
Mach One Trust | 1,194 | (3,230) | 50.0 | 4.16 |
Morgan Stanley Capital I* | 364 | (3,272) | 50.0 | 55.62 |
Morgan Stanley Capital I* | 1,542 | (3,458) | 22.3 | 23.00 |
JP Morgan Chase Comm Mtg Sec* | 934 | (3,868) | 42.0 | 3.58 |
Total | $ 7,451 | (27,602) |
* These securities are also included in the table on page 53 of our September 30, 2009 Form 10-Q, as they represent the securities with the five largest unrealized/unrecognized balances as of September 30, 2009.
We have been stress testing CMBS fixed maturity securities using conditional default rate assumptions of up to 2.5, which are above historic levels, to evaluate the duress under which a security will continue to perform and/or start to experience credit loss. If the above stress tests indicate a potential impairment of a particular security, we perform a discounted cash flow analysis on that particular security to ascertain whether or not the security is in fact other-than-temporarily impaired, as well as the amount of the credit impairment, if any. When performing the discounted cash flow analysis, we use the conditional default rates used during the stress testing of the security as a reference point, but we ultimately expand our testing using conditional default rates at more elevated levels in order to determine our best estimate of the present value of future cash flows.
The conditional default rate is the proportion of all loans outstanding in a security at the beginning of a time period that is expected to default during that period. Conditional default rate assumptions apply at the total collateral pool level held in the securitization trust. Generally, collateral conditional default rates will “ramp-up” over time as the collateral seasons, the performance begins to weaken and losses begin to surface. As time passes, depending on the collateral type and vintage, losses will peak and performance will begin to improve as weaker borrowers are removed from the pool through delinquency resolutions. In the later years of a collateral pool’s life, performance is generally materially better as the resulting favorable selection of the portfolio improves the overall quality and performance. While “ramped up” assumptions are used in our discounted cash flow analysis, when performing the stress testing of our CMBS portfolio, we typically apply a more conservative approach and do not apply a “ramp” of our conditional default rate assumptions, instead we assume the cash flows for the next period will experience defaults at the higher end of the range and then remain at that level for the life of the position, due to the current uncertainty surrounding the magnitude of potential future defaults on CMBS.
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 8
We believe the conditional default rates used are prudent as recent industry publications indicated a weighted average historical conditional default rate of approximately 0.9 for CMBS, which is based on vintage years dating back to the mid-1990s. In addition, our assumption of conditional default rates takes into consideration the uncertainty of future defaults as well as the fact that to date these securities have not experienced significant cumulative losses or delinquencies.
Consistent with our discounted cash flow analysis, our stress tests include a loan loss severity assumption, which is applied at the loan level of the collateral pool. The loan loss severity assumptions represent the estimated percentage loss on the loan-to-value exposure for a particular security. If the current loan-to-value ratio of a security was not available, we assumed a 50% loan loss severity. However, certain of the securities have lower current loan-to-value ratios, which in our opinion results in a severity assumption of 50% being overly conservative. Where we have current loan-to-value information and the loan-to-value ratio is lower than 80%, we adjust the severity assumption to reflect the fact that the loan-to-value ratio is lower.
For purposes of our stress tests, the loan loss severity assumption was held constant and was derived in either one of two ways:
(i) | Applying an estimated 60% loss on exposure percentage to the current loan-to-value ratio of a particular security (severity of 50% on a security with an 80% loan-to-value ratio equates to an estimated 60% loss on that loan-to-value exposure); or |
(ii) | Using an assumed 50% in those instances where current loan-to-value ratios were not available at the time of our assessment. |
We believe a 50% severity, and applying an estimated 60% loss on the loan-to-value exposure where loan-to-value is less than 80%, is appropriate as of September 30, 2009 as there has been an approximate 40% decrease in commercial real estate values from their peak. Our process resulted in a loan loss severity of between 42% - 50% for seven out of the eight securities, as compared to recent industry publications which indicated a weighted average loan loss severity under 35% recorded since 2002 on CMBS securities. For the one additional security that was not between 42% - 50%, Morgan Stanley, the loan loss severity of 22% reflects a current loan-to-value ratio of approximately 37%. The weighted average loan-to-value ratio on the eight securities was 66% based on their September 30, 2009 carrying values.
The “first loss” conditional default rate represents the default level at which the security will start to experience losses. The “first loss” conditional default rates on the eight securities listed above exceed our best estimate of defaults as used in our stress testing as of September 30, 2009. For both of the Morgan Stanley securities, the “first loss” conditional default rates exceeded our best estimate by a reasonably wide margin, which led us to the conclusion that these securities were not other-than-temporarily impaired. However, for the remainder of these securities, the margin between our best estimate used in stress testing and the ”first loss” conditional default rate was smaller and therefore we took into consideration other qualitative information relative to our conclusions on each of these securities. Those qualitative considerations were as follows for each of the remaining securities:
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 9
GS Mortgage Securities Corp (Total unrealized/unrecognized loss of $0.7 million)
This security is a subordinated tranche of a CMBS collateralized by a large and diverse (both by type and location) pool of commercial real estate loans. We have previously recognized several other-than-temporary impairment credit related charges on this security totaling $1.9 million. However, when consideration is given to the loan-to-value ratios (reported loan-to-value ratios are 73.6%) and the underlying properties’ ability to cover debt service (the current reported debt service coverage ratio is 1.2), we believe the amortized cost subsequent to previous other-than-temporary impairment charges was recoverable as of September 30, 2009.
ACT Depositor Corp. (Total unrealized/unrecognized loss of $2.6 million)
This security is a senior tranche of a Re-REMIC (a re-securitization) that represents an undivided interest in a static pool of seasoned subordinated CMBS. The underlying securities are well-seasoned with an average of seven years. Approximately 20% of the loans have been defeased and there is no longer any meaningful risk of default or loss. The current credit support of this senior tranche remains significant at approximately 70%; therefore, the collateral pool can realize material losses before this tranche would begin to experience an impairment.
GS Mortgage Securities Corp II (Total unrealized/unrecognized loss of $7.1 million)
This security is a senior tranche of a Re-REMIC (a re-securitization) that represents an undivided interest in a static pool of seasoned subordinated CMBS. These properties are well-seasoned at an average of six years. Approximately 13% of the loans have been defeased and therefore no longer carry any meaningful risk of default or loss. The current credit support of this senior tranche remains significant at approximately 16%; therefore, the collateral pool can realize material losses before this tranche would begin to experience an impairment.
JP Morgan Chase Comm MTG Sec (Total unrealized/unrecognized loss of $3.4 million)
This security is a senior tranche of a Re-REMIC (a re-securitization) that represents an undivided interest in a static pool of seasoned subordinated CMBS. These properties are well-seasoned at an average of four years and the quality of the underlying pool of securities, of which approximately 70% are investment grade, demonstrated ability to support the mortgages with a weighted average reported debt service coverage ratio of approximately 1.7, and modest advance rates. Subsequent to our stress testing, we received notification that the security has a weighted average reported loan-to-value of approximately 71%, which would in-turn have reduced the loan loss severity in our stress testing from 50% to approximately 43%. The current credit support of this senior tranche remains significant at approximately 18%; therefore, the collateral pool can realize material losses before this tranche would begin to experience an impairment.
Mach One Trust (Total unrealized/unrecognized loss of $3.2 million)
This security is a senior tranche of a Re-REMIC (a re-securitization) that represents an undivided interest in a static pool of seasoned subordinated CMBS. These properties are well-seasoned at four years. In addition, the current credit support of this senior tranche remains significant at approximately 27%; therefore, the collateral pool can realize material losses before this tranche would begin to experience an impairment.
Jim B. Rosenberg, Senior Assistant Chief Accountant
United States Securities and Exchange Commission
December 15, 2009
Page 10
JP Morgan Chase Comm MTG Sec (Total unrealized/unrecognized loss of $3.9 million)
This security is a subordinated tranche of a CMBS collateralized by a large and diverse (both by type and location) pool of commercial real estate loans. These properties demonstrated ability to service the underlying mortgages, with a reported debts service coverage ratio of approximately 1.7. The current credit support of this subordinated tranche remains adequate at approximately 5%; therefore, the collateral pool can realize losses before this tranche would begin to experience an impairment.
As a result of our analysis described above, coupled with the fact that we do not have the intent to sell these securities and do not believe we will be required to sell these securities, we have concluded that these securities were not other-than-temporarily impaired as of September 30, 2009.
The Company acknowledges that:
· | The Company 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 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. |
Should you have any questions on our responses to your comments above, please give me a call at (973) 948-1774.
Very truly yours,
/s/ Dale A. Thatcher
Dale A. Thatcher
Executive Vice President, Chief Financial Officer & Treasurer
Selective Insurance Group, Inc.
cc: | United States Securities and Exchange Commission: |
Sasha Parikh
Mark Brunhofer