January 31, 2011
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
Division of Corporation Finance
United States Securities and Exchange Commission
100 F Street, N.E.
Washington, D.C. 20549
| Form 10-K for the Fiscal Year Ended December 31, 2009 |
| Schedule 14A Filed April 12, 2010 |
| Form 10-Q for the Quarterly Period Ended September 30, 2010 |
Dear Mr. Rosenberg:
This letter responds to your letter to The PMI Group, Inc. (“PMI”) and its Chief Executive Officer, L. Stephen Smith, dated December 30, 2010. In your letter, you provided PMI with comments on its Form 10-K for the Fiscal Year Ended December 31, 2009, Schedule 14A filed April 12, 2010 and Form 10-Q for the Quarterly Period Ended September 30, 2010. We respectfully address your comments below. For your convenience, we have quoted your comment immediately prior to our response.
Form 10-K for the Fiscal Year Ended December 31, 2009
Item 7. Management’s Discussion and Analysis
Conditions and Trends Affecting our Business, page 74
| 1. | You disclose on page 2 that in 2009, loss mitigation efforts contributed to approximately $1 billion of workouts of delinquent risk in force. Please provide us draft disclosure for future filings to separately quantify, in terms of risk in force, your loss mitigation efforts related to each significant type of loss mitigation activity, including workouts and loan modifications. Also, describe in the draft disclosure each significant type of loss mitigation activity. In your draft disclosure, quantify the amount of risk in force related to loan modifications that resulted in reduced mortgage payments from interest rate |
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 2 of 33
| and/or amortization period adjustments, and separately, quantify the modifications that resulted in loan principal reductions. |
PMI Response:
As set out below, in PMI’s Annual Report on Form 10-K for the year ended December 31, 2010 and periodic filings thereafter as appropriate, PMI will include additional disclosure describing and quantifying its significant loss mitigation activities. We note that when servicers report loan modifications to PMI, they may report some, all or none of the specific changes in the loans’ terms. Because servicers’ reporting practices are inconsistent and often delayed, the information PMI has about a particular loan modification may be incomplete. As a result of these reporting issues and because loan modifications often include any combination of modified terms, we do not separately track our modified risk-in-force with respect to the individual categories such as interest rate reductions, amortization adjustments or loan principal reductions.
Proposed Disclosure:
Form 10-K, Item 1. Business
PMI’s Homeownership Preservation Initiatives department (“HPI”) was formed in 2007 to facilitate and enhance retention workouts on PMI-insured loans. Retention workouts include loan modifications and other loan repayment options discussed below, which may enable borrowers to cure mortgage defaults and retain ownership of their homes. HPI coordinates with loan servicers to provide them with delegated authority to approve workouts on certain categories of loans within specified parameters, places PMI loss mitigation staff on-site at loan servicers’ offices and trains servicer staff on loan workout strategies. HPI also supplements loan servicers’ outreach efforts to delinquent borrowers on PMI-insured loans to educate them about workout options and to facilitate loan workout discussions with their loan servicers. HPI provides these outreach efforts through the use of PMI’s loss mitigation professionals and through the engagement of specialty vendors and non-profit housing counseling agencies. If a retention workout is not viable for a borrower, PMI attempts to mitigate loss on the loan through a liquidation workout option, including a pre-foreclosure sale or a deed-in-lieu of foreclosure.
Retention workouts are designed to assist borrowers in curing, or satisfying in full, their delinquent loans. The ability of borrowers to cure delinquent loans is influenced by borrowers’ financial resources, regional housing and economic conditions, such as unemployment rates and home prices, and in recent periods, the speed and efficiency with which loan servicers process delinquencies and engage in loss mitigation efforts. A retention workout is often preceded by a forbearance plan, which temporarily allows the suspension of all or part of a borrower’s regularly scheduled payment for a specified time period to permit time to explore retention workouts. Retention workout options include refinances, loan modifications (Home Affordable Modification Program (“HAMP”) and non-HAMP), and repayment plans. With a loan refinance, the borrower obtains a new loan and uses the proceeds to satisfy in full the existing loan. Modification of a PMI-insured loan is a process
2
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 3 of 33
whereby any or some of the terms of a loan are modified with the approval of the insured lender and PMI. In a loan modification, the parties may agree to adjust the interest rate, extend the amortization period, and/or increase the amount of principal by the delinquent amount. Primarily because loan modifications often include any combination of modified terms, we do not separately track our modified risk-in-force with respect to the individual components of the loan modification. A loan repayment plan permits a delinquent borrower to satisfy its past-due loan amount by making additional payments along with regularly scheduled payments until the borrower becomes current on the loan. Because it results in a curing of the underlying default, when a retention workout is completed, we remove the loan from PMI’s default inventory and cease to include it in our loss reserve estimates. While a loan is in forbearance, the underlying default remains and we continue to include the loan in PMI’s default inventory.
A liquidation workout may be in the form of a pre-foreclosure sale (early disposal of the underlying property, usually in an amount less than what the borrower owes) or a deed-in-lieu of foreclosure (where the borrower conveys title of a property securing a loan to the lender in satisfaction of the debt), both of which avoid the lengthy and more costly foreclosure process. Both options can mitigate PMI’s loss by reducing foreclosure and other costs, resulting in a lower claim size to PMI.
Management’s Discussion and Analysis of Financial Condition and Results of Operations
Conditions and Trends Affecting our Business
U.S. Mortgage Insurance Operations
Defaults
Loss Mitigation Activities
The following tables show risk-in-force for each quarter that has been reported to us as being newly enrolled in forbearance plans as of each period end:
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| | Q3 2010 | | | Q2 2010 | | | Q1 2010 | | | Q4 2009 | | | Q3 2009 | | | Q2 2009 | | | Q1 2009 | |
| (Dollars in thousands) | |
New risk-in-force subject to forbearance in the quarter | | $ | 114,756 | | | $ | 177,672 | | | $ | 513,355 | | | $ | 656,093 | | | $ | 646,012 | | | $ | 198,723 | | | $ | 32,235 | |
Forbearance plans temporarily allow the suspension of all or part of a borrower’s regularly scheduled payment for a specified time period and include risk-in-force subject to HAMP and non-HAMP trial modification periods. After the introduction of the HAMP program in April 2009, as certain of our delinquent insured loans entered trial modification periods in each quarter, we experienced an increase in risk-in-force subject to forbearance plans. As a result of a decreasing population of HAMP eligible borrowers, the amount of new risk-in-force entering forbearance plans each quarter began to decline in the second half of 2010. As of September 30, 2010, approximately $2.4 billion of our risk-in-force was in a forbearance plan. Because a significant number of the loans subject to forbearance plans are in trial modification periods, we believe that they are an indicator of risk-in-force that could ultimately cure as a result of
3
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 4 of 33
retention workouts. Consequently, a portion of the loans subject to forbearance plans in a particular quarter may be re-characterized in a later period as risk related to a retention workout.
Retention workouts are designed to result in borrowers curing, or satisfying in full, their delinquent loans. Retention workouts include refinances, loan modifications and repayment plans. A liquidation workout is designed to mitigate PMI’s loss on a paid claim through options such as a pre-foreclosure sale or a deed-in-lieu of foreclosure. The following tables show mitigated risk resulting from retention and liquidation workouts completed in each of the quarters indicated:
| | | | | | | | | | | | | | | | | | | | |
| | | | | Q3 2010 | | | Q2 2010 | | | Q1 2010 | | | Total | |
| | | | | (Dollars in thousands) | |
Retention workouts | | | | | | $ | 462,727 | | | $ | 542,405 | | | $ | 543,035 | | | $ | 1,548,168 | |
Liquidation workouts | | | | | | $ | 140,115 | | | $ | 162,164 | | | $ | 149,542 | | | $ | 451,820 | |
| | | | | |
| | Q4 2009 | | | Q3 2009 | | | Q2 2009 | | | Q1 2009 | | | Total | |
| | (Dollars in thousands) | |
Retention workouts | | $ | 285,000 | | | $ | 204,388 | | | $ | 187,390 | | | $ | 212,604 | | | $ | 889,382 | |
Liquidation workouts | | $ | 125,894 | | | $ | 112,573 | | | $ | 115,404 | | | $ | 87,397 | | | $ | 441,268 | |
Note: Year-end aggregate numbers may not total due to rounding.
Beginning in the fourth quarter of 2009, we experienced a significant increase in mitigated risk-in-force as a result of retention workouts, which peaked in the first half of 2010 and has since declined. The increase in risk subject to retention workouts was primarily driven by loans completing their HAMP trial modification periods and loans returning to current status. Risk-in-force subject to retention workouts started to trend downward toward the end of 2010 as a result of a decline in our delinquent loan inventory and servicers refining their HAMP modification programs to require verification of income and other qualifying information during the initial stage of the modification process. Liquidation workouts of our delinquent risk have been elevated in 2009 and 2010 primarily as a result of the increased incidence of short sales and deeds-in-lieu approved by servicers and PMI as a means of mitigating loss by reducing foreclosure and other costs.
| 2. | You disclose on page 78 that past rescission activity has materially reduced your loss reserve estimates. Please provide us draft disclosure for future filings to include the following: |
4
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 5 of 33
| [a] | If true, it is your legal right to unilaterally rescind coverage on your mortgage insurance policies as if such coverage never existed, and disclose the conditions that must be met in order for this right to be exercised. |
| [b] | The rights that the policyholder or beneficiary has as it relates to premium refunds, including whether there are situations that result in a premium refund other than when a policy is rescinded by you. For example, would there be a right to a premium refund on a loan that does not meet underwriting criteria, but nevertheless, is not in default or never defaults? Please disclose in the Notes to your financial statements how you are accounting for such refunds, and tell us the applicable accounting literature upon which you relied. |
| [c] | You disclose on page 79 that you experienced an increase in the number of disagreements in 2009 with your decision to rescind coverage, and if you are unsuccessful in defending your rescission decisions beyond expected levels, you may need to re-establish loss reserves for such rescinded loans. Please disclose the amount of risk in force subject to these disagreements. |
| [d] | How, as an operational matter, a reinstatement works, including the contractual rights of the insured to appeal and/or arbitrate the rescission, the timing of the appeals and/or arbitration process, and at what point you deem a rescission is resolved. |
| [e] | Whether you have sought or are planning to seek recoveries from mortgage insurance beneficiaries for previously paid claims. If so, please disclose in the Notes to your financial statements how you are accounting for such recoveries, and tell us the applicable accounting literature upon which you relied. |
PMI Response:
As described below, in PMI’s Annual Report on Form 10-K for the year ended December 31, 2010 and periodic filings thereafter as appropriate, PMI will include additional disclosure regarding its rescission activity.
| [a] | Contractual Rescission Rights |
We provide additional disclosure describing PMI’s rescission rights and the process PMI follows when it rescinds coverage in paragraphs 2 through 3 of Appendix A.
5
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 6 of 33
We account for premium refunds in accordance with the terms of the applicable PMI mortgage insurance policy. Under PMI’s primary mortgage insurance master policy, which terms govern the significant majority of PMI’s risk-in-force, in the event coverage is rescinded for any loan, PMI refunds all premiums for that loan for the period following the occurrence of the event giving rise to the rescission. In nearly all cases, the event giving rise to rescission was at loan origination. Accordingly, when PMI rescinds insurance coverage, we refund all premiums associated with the insured loan, whether or not the loan was in default. In the absence of authoritative guidance, we follow Topic 944 by accounting for premium refunds as an offset to premiums earned, consistent with industry accounting practice and the AICPA Audit and Accounting Guide for Property and Liability Insurance Entities. As such, earned premiums are shown net of premium refunds on our Consolidated Statements of Operations. We provide additional disclosure to be included in Management’s Discussion and Analysis of Financial Condition and Results of Operations and Notes to Financial Statements regarding PMI’s treatment of premium refunds in paragraphs 2 and 7 of Appendix A.
| [c] | Rescission of Coverage Disagreements |
PMI’s insureds and their servicers have challenged, and are challenging, the bases of our rescissions of coverage in a variety of ways and with varying specificity. We receive requests from insureds and/or servicers for us to reconsider individual rescission decisions we have made on specific loans (“loan-level challenges”). We can quantify the amount of risk currently associated with loan-level challenges. We are also currently engaged in litigation or arbitration (or other alternative dispute resolution mechanisms) with respect to populations of loans for which we have rescinded coverage. We believe that we can reasonably estimate or quantify the amount of risk associated with such populations. We have also received challenges with respect to our general rights to rescind coverage of loans under the terms of PMI’s master policy (“general challenges”). These general challenges do not identify the specific loans believed to be at issue and may refer to broad categories of rescissions. Accordingly, we are unable to quantify the aggregate risk amount that may be subject to general challenges. Given this difficulty, PMI proposes to provide additional tabular disclosure showing, for each year in which the risk was written, the aggregate delinquent and non-delinquent rescinded risk to date and the aggregate risk amount of rescinded loans which PMI believes are now subject to disagreement either because PMI has received a loan-level challenge or because the loan is subject to litigation, arbitration or other dispute resolution process. Please see paragraphs 4 and 5, and Table 2, in Appendix A for our proposed disclosure. We also
6
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 7 of 33
provide narrative disclosure with respect to general challenges that we have received in paragraph 6 of Appendix A.
We provide additional disclosure regarding lender requests for reconsideration of rescinded loans and reinstatements in paragraphs 4 through 6, and Table 2, of Appendix A.
After PMI pays a claim, we generally do not seek to recover claim payments from our insureds. The limited instances in which we may seek reimbursement of a claim payment include our discovery that an insured has erroneously advised PMI that a condition precedent for claim payment had been fulfilled. When we recover a claim payment, we record it as a reduction to our incurred losses at the time of receipt. Because our claim recoveries, both in number and in aggregate dollar amounts, have not been material, we have not discussed their accounting treatment in our Notes to Financial Statements. In the event that recoveries become a material component of our business, we would so disclose and discuss the underlying accounting and reimbursement processes and recovery amounts.
Results of Operations
Segment Results
U.S. Mortgage Insurance Operations
Insurance and risk in force, page 91
| 3. | Please provide us draft disclosure for future filings to revise your tabulardisclosure of the ten states with the highest risk in force shown on page 91 and the loans with certain characteristics identified by PMI as having heightened risk (per page 26) shown on page 93 to also disclose the associated reserve for losses and loss adjustment expenses related to each line item. |
PMI Response:
We do not estimate loss reserves on a loan-by-loan basis. While we take into consideration the geographic mix of our delinquent loan inventory and loan and risk characteristics within the inventory, these are only some of the several factors we consider in setting loss reserves. (See the discussion of our loss reserve practices and methodology in Appendix B.) Accordingly, while we can provide disclosure showing our reserves associated by geography and loan characteristics, we believe that it should be made clear that such data is not a reflection or indication of how those reserves were established. With this in mind, we propose the following
7
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 8 of 33
additional disclosure in PMI’s Annual Report on Form 10-K for the year ended December 31, 2010 and periodic filings thereafter as appropriate.
Proposed Disclosure:
The table below sets forth the percentage of PMI’s primary risk-in-force as of September 30, 2010 and December 31, 2009 in the ten states with the highest risk-in-force as of September 30, 2010 in PMI’s primary portfolio and the reserve for losses and loss adjustment expenses associated with delinquent risk-in-force in those states. The geographic mix of our delinquent loan inventory is one of many factors we consider when estimating loss reserves. While we do not determine loss reserves based on a geographic location, the table below reflects an allocation of loss reserves based on such information. For a discussion of our loss reserve estimation process, seeManagement’s Discussion and Analysis of Financial Condition and Results of Operations - Critical Accounting Estimates – Reserves for Losses and LAE.
| | | | | | | | | | | | | | | | | | | | | | | | |
| | As of September 30, 2010 | | | As of December 31, 2009 | |
| | Percentage of Primary Risk in Force | | | Reserve for Losses and LAE | | | Reserves as a Percentage of Total USMI Reserves | | | Percentage of Primary Risk in Force | | | Reserve for Losses and LAE | | | Reserves as a Percentage of Total USMI Reserves | |
| | (Dollars in thousands) | | | (Dollars in thousands) | |
Florida | | | 9.70 | % | | $ | 585,677 | | | | 19.6 | % | | | 10.10 | % | | $ | 617,995 | | | | 19.2 | % |
California | | | 7.40 | % | | | 333,314 | | | | 11.2 | % | | | 7.70 | % | | | 397,216 | | | | 12.4 | % |
Texas | | | 7.90 | % | | | 102,707 | | | | 3.4 | % | | | 7.50 | % | | | 108,020 | | | | 3.4 | % |
Illinois | | | 5.20 | % | | | 170,958 | | | | 5.7 | % | | | 5.20 | % | | | 164,458 | | | | 5.1 | % |
Georgia | | | 4.70 | % | | | 119,616 | | | | 4.0 | % | | | 4.70 | % | | | 128,467 | | | | 4.0 | % |
New York | | | 4.00 | % | | | 115,705 | | | | 3.9 | % | | | 4.00 | % | | | 103,417 | | | | 3.2 | % |
Ohio | | | 3.90 | % | | | 69,898 | | | | 2.3 | % | | | 3.90 | % | | | 70,946 | | | | 2.2 | % |
Pennsylvania | | | 3.50 | % | | | 57,334 | | | | 1.9 | % | | | 3.40 | % | | | 54,794 | | | | 1.7 | % |
New Jersey | | | 3.30 | % | | | 106,737 | | | | 3.6 | % | | | 3.30 | % | | | 99,225 | | | | 3.1 | % |
Washington | | | 3.30 | % | | | 72,129 | | | | 2.4 | % | | | 3.20 | % | | | 64,513 | | | | 2.0 | % |
| | | | | | | | | | | | | | | | | | | | | | | | |
Total | | | | | | $ | 1,734,075 | | | | 58.1 | % | | | | | | $ | 1,809,051 | | | | 56.3 | % |
| | | | | | | | | | | | | | | | | | | | | | | | |
The table below presents PMI’s less-than-A quality, Alt-A, ARM, Above-97, interest only and payment option ARM loans as a percentage of primary risk-in-force and the associated reserve for losses and loss adjustment expenses associated with delinquent risk-in-force for each loan type. Loan and risk characteristics within PMI’s delinquent loan inventory are two of many factors we consider when estimating loss reserves. While we do not establish loss reserves based on these factors, the loss reserve amounts set out below have been allocated based on total delinquencies and also on the number of delinquencies associated with each risk characteristic. For a discussion of our loss reserve estimation process, seeManagement’s Discussion and
8
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 9 of 33
Analysis of Financial Condition and Results of Operations – Critical Accounting Estimates – Reserves for Losses and LAE.
| | | | | | | | | | | | | | | | | | | | | | | | |
| | As of September 30, 2010 | | | As of December 31, 2009 | |
| | As a Percentage of Primary Risk in Force | | | Reserve for Losses and LAE | | | Reserves as a Percentage of Total USMI Reserves | | | As a Percentage of Primary Risk in Force | | | Reserve for Losses and LAE | | | Reserves as a Percentage of Total USMI Reserves | |
Loan Type*: | | (Dollars in thousands) | | | (Dollars in thousands) | |
Less-than-A Quality loans (FICO scores below 620) | | | 6.60 | % | | $ | 291,214 | | | | 9.8 | % | | | 6.80 | % | | $ | 315,281 | | | | 9.8 | % |
Less-than-A Quality loans with FICO scores below 575** | | | 1.70 | % | | | 87,749 | | | | 2.9 | % | | | 1.70 | % | | | 95,983 | | | | 3.0 | % |
Alt-A loans | | | 15.70 | % | | | 938,710 | | | | 31.5 | % | | | 17.00 | % | | | 1,086,736 | | | | 33.8 | % |
ARMs (excluding 2/28 Hybrid ARMs) | | | 6.80 | % | | | 414,312 | | | | 13.9 | % | | | 7.50 | % | | | 473,465 | | | | 14.7 | % |
2/28 Hybrid ARMs*** | | | 1.40 | % | | | 124,622 | | | | 4.2 | % | | | 1.80 | % | | | 151,889 | | | | 4.7 | % |
Above-97s (Above 97% LTV’s) | | | 19.90 | % | | | 682,607 | | | | 22.9 | % | | | 20.40 | % | | | 741,659 | | | | 23.1 | % |
Interest Only | | | 10.10 | % | | | 541,255 | | | | 18.1 | % | | | 10.90 | % | | | 617,464 | | | | 19.2 | % |
Payment Option ARMs | | | 2.70 | % | | | 198,575 | | | | 6.7 | % | | | 3.10 | % | | | 233,413 | | | | 7.3 | % |
* | Loan types are not mutually exclusive as certain loans may be included in one or more of the above loan categories. The total reserve for losses and loss adjustment expenses associated with all loan types listed above was $2.8 billion and $2.9 billion for the periods ended September 30, 2010 and December 31, 2009 respectively. Total reserves for losses and loss adjustment expenses for the U.S. Mortgage Insurance Operations’ segment were $3.0 billion and $3.2 billion for the periods ended September 30, 2010 and December 31, 2009, respectively. |
** | Less-than-A quality loans with FICO scores below 575 is a subset of PMI’s less-than-A quality loan portfolio. |
*** | 2/28 Hybrid ARMs are loans whose interest rate is fixed for an initial two-year period and floats thereafter. |
Critical Accounting Estimates, page 113
| 4. | Your reinsurance recoverables increased by $220.87 million to $703.55 million as of December 31, 2009 compared to December 31, 2008, and your recoverables were primarily from captive arrangements, per Note 11, that primarily provide for excess-of-loss reinsurance, per page 31. Within Note 2, you disclose that reinsurance agreements typically provide for a recovery of a proportionate level of claim expenses from reinsurers; however, it appears that reinsurance recoveries related to excess-of-loss agreements would not be proportionate to the level of claim expenses. Please provide us draft disclosure for future filings to revise your accounting policy for reinsurance within your Notes to address excess-of-loss contracts, and expand your discussion within Critical Accounting Estimates to address the following related to your reinsurance recoverables estimate: |
| • | | The amount of reinsurance recoverable related to each type of reinsurance contract; |
9
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 10 of 33
| • | | The methodology and key assumptions you use to estimate your reinsurance recoverables; and |
| • | | Whether your reinsurance recoverables valuation methodology is consistent with the methodology you use for estimating reserves for losses and loss adjustment expenses. |
PMI Response:
In PMI’s Annual Report on Form 10-K for the year ended December 31, 2010 and periodic filings thereafter, PMI will include additional disclosure regarding our accounting policy for excess-of-loss reinsurance and our reinsurance recoverables estimate.
Proposed Disclosure:
Notes to Consolidated Financial Statements
Note 2. Summary of Significant Accounting Policies
Reinsurance— The Company uses reinsurance to reduce net risk-in-force and enhance capital allocation. This is done primarily with captive reinsurance companies via both excess-of-loss (“XOL”) and quota share arrangements. Under a captive reinsurance agreement, the Company reinsures a portion of its risk written on loans originated by a certain lender with the captive reinsurance company affiliated with such lender. PMI’s captive reinsurance agreements primarily provide for XOL reinsurance, in which PMI retains a first loss position on a defined set of mortgage insurance risk, reinsures a second loss layer of this risk with the captive reinsurance company and retains the remaining risk above the second loss layer. The Company cedes premiums pursuant to its captive reinsurance arrangements in exchange for the reinsurance of a portion of the Company’s risk less, in some instances, a ceding commission paid to us for underwriting and administering the business. Ceded premiums reduce premiums earned and ceded loss reserves reduce losses and loss adjustment expenses in the Consolidated Statements of Operations. Total loss reserves gross of reinsurance contracts are determined in accordance with the Company’s loss reserving practices. After gross reserves are established, we allocate reserves to reinsurance contracts consistent with the terms of the coverage provided and based on an actuarial analysis of the loans reinsured under each reinsurance agreement. The total ceded loss reserves subject to reinsurance arrangements represent reinsurance recoverables in the Consolidated Balance Sheets. Under the Company’s XOL captive reinsurance arrangements, PMI generally cedes 100% of the losses within the reinsured loss layer to the captive insurance company. PMI limits its recorded recoverable to the amount currently available in the trust account maintained by the captive insurance company for PMI’s benefit, if that amount is less than PMI’s ceded loss reserves. Effective January 1, 2009, the Company ceased seeking reinsurance under XOL captive reinsurance agreements. On that date, in-force XOL contracts were placed into run-off
10
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 11 of 33
and will mature pursuant to the existing terms and conditions. A small percentage of the Company’s existing captive reinsurance arrangements are under quota share agreements under which we continue to reinsure new insurance business.
Management’s Discussion and Analysis of Financial Condition and Results of
Operations Critical Accounting Estimates
U.S. Mortgage Insurance Operations
The following table shows amounts of reinsurance recoverables related to each type of reinsurance contract:
| | | | | | | | | | | | | | | | | | | | | | | | |
| | September 30, 2010 | | | December 31, 2009 | | | December 31, 2008 | |
| | Dollar Amount | | | Percentage | | | Dollar Amount | | | Percentage | | | Dollar Amount | | | Percentage | |
| | (Dollars in thousands) | | | (Dollars in thousands) | | | (Dollars in thousands) | |
Captives(1): | | | | | | | | | | | | | | | | | | | | | | | | |
Excess-of-loss | | $ | 519,079 | | | | 99.4 | % | | $ | 700,308 | | | | 99.5 | % | | $ | 480,251 | | | | 99.5 | % |
Quota share | | | 3,037 | | | | 0.6 | % | | | 3,241 | | | | 0.5 | % | | | 2,415 | | | | 0.5 | % |
Other | | | — | | | | 0.0 | % | | | — | | | | 0.0 | % | | | 12 | | | | 0.0 | % |
| | | | | | | | | | | | | | | | | | | | | | | | |
Total | | $ | 522,116 | | | | 100.0 | % | | $ | 703,550 | | | | 100.0 | % | | $ | 482,678 | | | | 100.0 | % |
| | | | | | | | | | | | | | | | | | | | | | | | |
(1) Amounts shown have been reduced by a valuation allowance on reinsurance recoverables, to the extent applicable, if they are in excess of captive trust account balances. The table below shows the total of all balances in trust accounts for the benefit of PMI. While certain recoverables from captives equal the amount in the respective captive trust account, the majority of captives have assets in excess of PMI’s related recoverable. | |
| | | | | September 30, 2010 | | | | | | December 31, 2009 | | | | | | December 31, 2008 | |
| | | | | | | | (Dollars in thousands) | | | | |
Off Balance Sheet Captive Trust Assets | | | | | | $ | 833,103 | | | | | | | $ | 940,664 | | | | | | | $ | 860,118 | |
| | | | | | | | | | | | | | | | | | | | | | | | |
Loss reserves ceded to captive reinsurers are recorded as reinsurance recoverables and reduce total incurred losses in our Consolidated Statements of Operations. Reinsurance recoverables are based on our estimates of unpaid claims as applied to the terms of the respective reinsurance arrangements. The methodology and key assumptions used in our estimate of reinsurance recoverables are the same as those used in our estimate of loss reserves and are allocated to reinsurance contracts using actuarial analysis for applicable reinsurance agreements. (For a discussion of our loss reserve estimation process and methodology, seeManagement’s Discussion and Analysis of Financial Condition and Results of Operations – Critical Accounting Estimates – Reserves for Losses and LAE.) As reinsurance recoverables are derived from our loss estimation process, amounts we will ultimately recover could differ materially from amounts recorded as reinsurance recoverables. PMI limits its recorded reinsurance recoverable to the amount currently available in the trust account maintained by the captive insurance company for PMI’s benefit, if that amount is less than PMI’s ceded loss reserves.
11
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 12 of 33
Reserves for Losses and LAE
U.S. Mortgage Insurance Operations, page 113
| 5. | You disclose on page 113 that you consider the effect of projected future rescission activity with respect to the current inventory of delinquent loans, and such consideration has materially reduced your loss reserve estimates, and to the extent you must reverse [sic] your rescissions beyond recent levels or future rescission activity is lower than expected, you may need to significantly increase your loss reserve estimates in future periods. Additionally, on page 73 of your Form 10-Q for the Quarterly Period Ended September 30, 2010, you disclose that your expectations regarding future levels of loan modifications, rescissions and claim denials factor into your loss reserve estimation process. Please address the following: |
| [a] | Tell us why it is appropriate under GAAP to reduce your loss reserves by estimating rescissions and claim denials. Please reference the authoritative accounting guidance upon which you relied. In addition, provide us draft disclosure for future filings to describe the methods you used to determine your estimate of loan modifications, rescissions and claim denials. |
| [b] | Please provide us draft disclosure for future filings to disclose the range of loss reserve estimates as determined by your actuaries. Discuss the key assumptions used to arrive at management’s best estimate of loss reserves within that range and what specific factors led management to believe this amount rather than any other amount within the range represented the best estimate of incurred losses. |
| [c] | You disclose on page 73 of your Form 10-Q for the Quarterly Period Ended September 30, 2010 that management’s best estimate was below the actuarial midpoint primarily due to your views with respect to the ultimate cure rates of your pending delinquencies, and that management believes that the likely ultimate benefit of the resolution of these delinquencies, including through loan modifications, loan work-outs and borrowers bringing their loans current, is not fully reflected in available data, and accordingly was not able to be fully considered in the development of the actuarial range. Please provide us draft disclosure for future filings to address the differences between the data that was available to the actuaries and the data that management utilized to make their best estimate. |
12
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 13 of 33
| [d] | Provide us draft disclosure for future filings as to the extent to which a backlog related to the processing of rescissions and reinstatements exists, and how the backlog affects your reserve estimate. |
| [e] | Provide us draft disclosure for future filings to indicate whether your reserve estimate includes an assumption for reinstatement of previously rescinded policies, and if not, your basis for not including such an assumption. If included, please disclose how you developed the assumptions used and quantify the amount for the periods presented. |
PMI Response:
| [a] | Loss Reserve Methodology |
We establish PMI’s reserves for losses and loss adjustment expenses (“LAE”) based upon our estimate of unpaid losses and LAE on (i) reported mortgage loans in default, and (ii) estimated defaults incurred but not reported to PMI by its customers. Our loss reserve methodology, consistent with industry practice, is based on Accounting Standards Codification (“ASC”) Topic 944, which states “the estimated liability for unpaid claims includes the amount of money that will be required for future payments on both of the following:
| a. | Claims that have been reported to the insurer |
| b. | Claims relating to insured events that have occurred but have not been reported to the insurer as of the date the liability is estimated.” |
ASC Topic 944 also notes that “the liability for unpaid claims shall be based on the estimated ultimate cost of settling the claims (including the effects of inflation and other societal and economic factors), using past experience adjusted for current trends, and any other factors that would modify past experience.”
Two key assumptions used in estimating the dollar amount to reserve on loans in our existing delinquent loan inventory are the frequency of claims (i.e., the number of delinquent mortgages that will ultimately result in a payment made by us, or the “claim rate”) and the size of the claims. To estimate the number or frequency of claims that we will pay, we estimate how many delinquencies within our current delinquency inventory will ultimately not result in a claim, or “cure”. The number of cures is a function of a number of possible outcomes, including borrowers bringing their loans current, selling their homes, or refinancing or modifying their loans, or PMI’s rescission of coverage of delinquent loans. To estimate the probable size of future paid claims, we evaluate, among other factors, PMI’s likely claim denial activity with respect to the current inventory of delinquent loans. Claim denials impact claim sizes because we include them
13
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 14 of 33
as zero dollar claim payments, which reduce our historical and future estimated average paid claim amounts. Because they have a material effect on PMI’s claim rates and claim size, we consider our future loan modification, rescission and claim denial activity in our estimation of the amount of money that will be required for future claim payments for existing delinquent loans. Our proposed disclosure in paragraphs 6 through 15 of Appendix B discusses how we estimate future loan modification, rescission and claim denial activity.
| [b] | Actuarial and Management Loss Reserve Estimates |
In paragraphs 1 through 7 of Appendix B, we provide proposed additional disclosure to be included in our Management’s Discussion and Analysis of Financial Condition and Results of Operations regarding the range of loss reserve estimates as determined by our actuaries, the key assumptions used to arrive at management’s best estimate of loss reserves within that range and the factors that led management to believe this amount within the range represented its best estimate.
| [c] | Differences in Data Available to, and Considered by, Actuaries and Management |
In paragraph 7 of Appendix B, we provide proposed additional disclosure to be included in our Management’s Discussion and Analysis of Financial Condition and Results of Operations regarding the differences in data available to, and considered by, our actuaries and management in the respective loss reserve analyses.
In paragraph 11 of Appendix B, we provide proposed additional disclosure describing the backlog in reviewing investigated loans and requests for reconsideration and the effect on our loss reserve estimates.
| [e] | Reinstatement Reserving Assumptions |
In paragraphs 9 and 11 of Appendix B, we provide proposed additional disclosure regarding management’s assumptions for reinstatement of previously rescinded policies.
14
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 15 of 33
| 6. | You disclose that your estimation process uses generally recognized actuarial projection methodologies on page 114. Provide us draft disclosure for future filings to specifically identify what methodologies and/or models you utilize. |
PMI Response:
In PMI’s Annual Report on Form 10-K for the year ended December 31, 2010 and periodic filings thereafter as appropriate, PMI will include additional disclosure identifying the generally recognized actuarial methodologies we utilize. Please see paragraphs 4 through 5 of Appendix B for our proposed disclosure.
| 7. | Refer to your disclosure on pages 114 and 115 and Note 8 related to the reconciliation of your U.S. Mortgage Insurance Operations’ beginning and ending reserves for losses and LAE, and provide us draft disclosure for future filings to address the following: |
| [a] | Quantify the change in the current year and prior year liability related to rescission activity, claim denials, modifications, and other significant loss mitigation efforts. Additionally, disclose the reduction in loss reserves separately related to rescissions and to loan modifications as of the end of each period presented; |
| [b] | Separately, quantify the adverse development in the modified pool insurance portfolio related to the prior year liability, including the offsetting effects of restructurings completed in 2009, as described on page 78; and |
PMI Response:
| [a] | Changes in Current Year and Prior Year Liability; Quantification of Reductions |
in Loss Reserves Related to Rescissions and Loan Modifications
When we establish loss reserves, we allocate separate portions between our primary and pool delinquent risk. The key factors used by management in establishing reserves for our primary and pool portfolios are claim rates and claim sizes. Management considers and refines these and other factors each reporting period. The net effect of these period over period refinements is a positive or negative change to loss reserves, allocated between delinquencies reported during the current calendar year and those reported in prior periods, and reflected in the tables contained in paragraphs 16 and 18 of Appendix B. In any given period, changes to the estimated claim rate are determined using an aggregate approach using both qualitative and quantitative factors. Among other factors,
15
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 16 of 33
management’s estimates of future claim rates are influenced by the estimated impact of future rescission, loan modification and other loss mitigation activity. Our projection of an average claim size is impacted in part by our claim denial activity. While management considers these factors when establishing loss reserves, it does not separately quantify each of their expected distinct effects on changes in reserves related to the current year and prior years’ liability in each reporting period. This is the case due in large part to the significant inter-relation of our assumptions and estimates and to management’s use of both qualitative and quantitative information and considerations. Accordingly, we propose additional disclosure, as set out in paragraphs 16 through 18 of Appendix B, to separately quantify the net effect of changes in the current year and prior period reserves between primary and pool portfolios, with qualitative discussion of the underlying drivers of any changes.
In paragraph 8 of Appendix B, we discuss our methodology with respect to projecting future modifications of delinquent insured loans. We state that management uses its estimate of future modifications to independently evaluate our actuarially developed claim rate and, ultimately, to set appropriate reserve levels. In the course of this analysis, management does not, however, attempt to isolate the explicit effect of modifications within selected claim rates or the dollar impact of any resulting benefit to our reserve balance. One of the primary reasons we do not attempt this modeling is that such an effort would require the estimation of the number of modified loans that would have cured even without their modification. We believe that such estimation would be highly speculative. To provide investors with additional quantitative information with respect to the impact of projected future loan modifications on our loss reserve balance, we propose to disclose management’s estimate of the percentage of our primary delinquent loan inventory that will cure due to future loan modifications and the estimated risk in force represented by such percentage. This proposed disclosure is set out in paragraph 8 of Appendix B.
We have recently enhanced our internal models with respect to future estimated rescissions and claim denials. Unlike loan modifications discussed above, we believe that we can provide investors with estimations of the benefits of future rescission and claim denial activity upon our loss reserve balances as of December 31, 2010 and thereafter. Therefore, for the period ended December 31, 2010 and thereafter, we propose to quantify the estimated reductions in loss reserves attributable to our projected future rescission and claim denial activity. Our proposed new disclosure, to be included in our Annual Report on Form 10-K for the year ended December 31, 2010 and periodic filings thereafter, is set out in paragraphs 9 and 15 of Appendix B.
| [b] | Quantify the Adverse Development in Modified Pool Insurance Portfolio |
As set out in paragraphs 16 through 18 of Appendix B, we propose to quantify in future filings the adverse development in both our primary and pool insurance portfolios,
16
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 17 of 33
including the effects of the restructurings completed in 2009 related to our modified pool contracts.
Premium Deficiency Analysis, page 122
| 8. | You state that there was no premium deficiency in your U.S. Mortgage Insurance Operations segment despite continued significant losses in 2009. Provide us draft disclosure for future filings of your expected premium excess for this segment. Refer to ASC 944 60-25-4. In addition, please tell us what consideration was given to separately analyzing the contracts, by channel or by product, within your U.S. Mortgage Insurance Operations for a premium deficiency. |
PMI Response:
Our quarterly premium deficiency analysis is performed in accordance with ASC 944-60-25-4. To date, it has not indicated the need to establish a premium deficiency reserve. As discussed below, we perform the analysis on a total portfolio basis. Historically, the later years of a mortgage insurance policy are profitable as losses tend to occur within the first several years of a contract. Accordingly the likelihood of a premium deficiency reserve being required has declined over time. In addition, as a result of their high credit quality, PMI’s more recent book years are expected to be very profitable and are contributing to the decreasing likelihood that we will have a premium deficiency reserve in the future.
In our SEC periodic filings each quarter, we have disclosed (i) that we conducted a premium deficiency analysis, (ii) our approach, including whether investment income is considered, (iii) the conclusion of the analysis, and (iv) the risk that continued losses could cause us to establish a premium deficiency in the future. We believe that this disclosure satisfies the guidance and related disclosure requirement of the accounting literature. The determination of whether a premium deficiency exists is a binary measure, and we are not aware of any requirement to disclose the amount by which any test (e.g., premium deficiency analysis, impairment, etc.) exceeds the threshold for recognition of a reserve or impairment charge. Accordingly, we respectfully suggest that such disclosure not be provided.
With regard to how we analyze our various types of insurance contracts, we primarily acquire and service contracts and measure profitability on an over-all portfolio basis. Therefore, we perform our premium deficiency analysis based on a “single book” concept within each of our
17
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 18 of 33
operating segments. As set out in our comment response letter to you of May 27, 2008, with respect to our U.S. Mortgage Insurance Operations, our largest operating segment, the “single book” concept is supported by the following factors:
• | | We predominately issue all mortgage insurance policies in the U.S. through one insurance company: PMI Mortgage Insurance Co. (“MIC”). |
• | | MIC is a monoline property and casualty insurer and is only licensed to write insurance policies on first loss single family residential mortgage guaranty risk. |
• | | MIC’s mortgage insurance policy terms do not materially vary by product or channel. |
• | | Policies in the flow and structured channels are serviced in the same platform and in the same manner. |
• | | Subsequent to origination, risks inherent in the portfolio as well as profitability are measured and managed on an overall portfolio basis. |
• | | Multiple portfolio attributes or risks such as Interest Only, ARMs, Low/High LTV, etc. can be common to individual loans, which is why we manage our business on a single book basis. |
In summary, while we maintain data on certain characteristics of our policies in force (e.g., channel type, single-coverage or pool, and loan type), we service, manage and measure the profitability of our in force business on an overall portfolio basis within our operating segments. As a result, to be consistent with how we manage our business, we evaluate our in force insurance on a “single book” basis for each operating segment to assess the need for a premium deficiency reserve, which we believe to be the most appropriate approach given the nature of our operations and characteristics outlined above. Consistent with the above, we have performed our overall premium deficiency analysis for all U.S. book years combined and determined that a premium deficiency reserve was not required.
Schedule 14A Filed April 12, 2010
Executive Compensation
2009 Long-Term Cash Incentives, page 25
| 9. | We note your statement that the specific numeric metrics for Cash Components A and B are confidential and that the release of such information is likely to cause competitive harm to the Company. Please confirm that when you measure the achievement of each Component, at the end of the two and three-year performance periods, you will disclose the specific numeric metrics that are now omitted and the results corresponding to the level actually achieved. We note that the metrics are the type of information you regularly disclose on a historical basis. |
18
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 19 of 33
PMI Response:
We confirm.
Grants of Plan-Based Awards in 2009, page 34
| 10. | The column “Grant Date Fair Value of Stock and Options Awards ($)” should disclose the aggregate grant date fair value of each stock and option award rather than the value per stock unit or per stock option. Please confirm that in future filings you will disclose aggregate grant date fair values. |
PMI Response:
We confirm.
Form 10-Q for the Quarterly Period Ended September 30, 2010
Notes to Consolidated Financial Statements
Note 7, Reserve for Losses and Loss Adjustment Expenses (LAE), page 23
| 11. | You disclose the increase in prior years’ reserves in the first nine months of 2010 was partially offset by the recognition of a future cash flow stream with an estimated fair value of $82.3 million, as of the restructure dates, related to your restructuring of modified pool policies in the first and second quarter of 2010. Provide us draft disclosure for future filings to state the nature of this future cash flow stream, including your right to the cash flow stream and how you estimated its fair value. |
PMI Response:
We recognized a future cash flow stream with respect to the accounting for several particular modified pool restructuring transactions we completed in 2010. As a result of these restructuring transactions, we completely fulfilled our obligations under the applicable insurance contracts and determined that the restructurings constituted an extinguishment of the underlying modified pool policies. However, pursuant to the terms of the restructurings, we retained the right to the mortgage insurance premium payments from the underlying mortgages. In measuring the fair value of the consideration exchanged, we considered market valuations of instruments comparable to our rights to these future payments from these assets.
We believe that this stream of cash flows is similar to an interest-only strip from a GSE-backed securitization. To determine the fair value at initial recognition, we consulted an active participant in the GSE-backed securitization market. We obtained the pricing for interest-only
19
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 20 of 33
strips of securitizations with collateral that has similar characteristics to those of the pools of loans underlying our cash flow streams. Utilizing this market data, we calculated our estimate of fair value. We also determined a market discount rate for future interest accretion in this process.
To account for this restructuring, management released the accrued loss reserves and recognized the fair value of the cash flow stream. The excess of the loss reserve release and cash flow stream over the cash payment was recognized as a reduction in loss expense in our consolidated statement of operations. The fair value of the cash flow stream is recognized in other receivables on our consolidated balance sheet. Subsequently, we record accretion income in the other income (loss) line item on our consolidated statement of operations.
In PMI’s Annual Report on Form 10-K for the year ended December 31, 2010 and periodic filings thereafter as appropriate, PMI will include additional disclosure regarding the accounting treatment of the future cashflow stream related to certain modified pool restructurings.
Proposed Disclosure:
Management’s Discussion and Analysis of Financial Condition and Results of Operations
Results of Operations
Consolidated Results
In 2010, we restructured various modified pool policies whereby we paid a fee to be relieved of all of our remaining obligations to provide insurance coverage under those policies. In these transactions, we retained the right to continue to receive monthly premium payments as though the insurance had not been cancelled. We have determined that this stream of payments, which is supported by an underlying pool of mortgage loans, is no longer part of an insurance contract and instead must be accounted for separately as a discounted receivable. Management determined that this stream of cash flows most resembles a beneficial interest from a residential mortgage securitization (interest-only strips), which is also a stream of cash flows based on an underlying pool of mortgages.
To initially value this asset, management used observable market inputs from an active market with respect to interest-only strips for mortgage securitizations with similar characteristics to those of the former modified pool policies (e.g., origination year, loan-to-value ratios, etc.) and applied these fair value inputs to our expected cash flows. Consistent with ASC 835Interest, a market discount rate was calculated based on the excess of all cash flows attributable to the beneficial interest estimated at the acquisition date over the initial investment. On a monthly basis, we apply the cash received from this asset to the accretion with the excess cash reducing the value of the asset. The income recognition from the accretion yield is recognized as other income (loss).
* * *
20
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 21 of 33
PMI acknowledges that:
| • | | It 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 |
| • | | It 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. |
Thank you for your attention to this letter and for providing PMI with an opportunity to address the Staff’s comments. Should you have any questions, please feel free to contact me at (925) 658-6530 ordon.lofe@pmigroup.com.
|
Sincerely, |
|
/s/ Donald P. Lofe, Jr. |
|
Donald P. Lofe, Jr. |
Executive Vice President, Chief Financial Officer and |
Chief Administrative Officer |
The PMI Group, Inc. |
cc: | Staci Shannon, Staff Accountant |
| Joel Parker, Accounting Branch Chief |
| Brian Pitko, Staff Attorney |
| Sebastian Gomez Abero, Staff Attorney |
| (United States Securities and Exchange Commission |
John L. Savva
(Sullivan & Cromwell LLP)
21
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 22 of 33
Appendix A
Existing and Proposed Rescission Disclosure1
Management’s Discussion and Analysis of Financial Condition and Results of Operations
Conditions and Trends Affecting our Business
1. PMI routinely investigates early payment default loans (loans that default prior to the thirteenth payment), or EPDs, and also investigates certain other non-EPD loans, for misrepresentation, negligent underwriting and eligibility for coverage. Based upon PMI’s investigations, industry data and other data, we believe that there were significantly higher levels of mortgage origination fraud and decreases in the quality of mortgage origination underwriting in 2006 and 2007 when compared to historical levels. As a result, PMI continues to review and investigate a substantial volume of insured loans.
2.PMI’s mortgage insurance policies, certain endorsements, and certain of its lender-paid mortgage insurance commitment agreements contain provisions giving PMI the right to unilaterally rescind coverage of an insured loan for breach of representations and warranties, material misrepresentation, negligent underwriting and/or ineligibility. PMI may also have rescission rights under general principles of contract law. Generally, PMI exercises its contractual rights following an investigation and upon establishing a reasonable belief that at loan origination there was a material misrepresentation by the originator or an agent of the originator, that the loan was negligently underwritten, or that the loan was not eligible for coverage under an approved loan program or set of underwriting guidelines. When PMI rescinds coverage, we notify the insured in writing, identify the bases for the rescission, summarize the evidence supporting the rescission decision and refund all premiums associated with the rescinded loan to the insured, whether or not the loan was delinquent.The tables below show the aggregate risk of delinquent and non-delinquent loans rescinded by PMI in each quarter in 2010 and 2009:
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| | Q3 2010 | | | Q2 2010 | | | Q1 2010 | | | Q4 2009 | | | Q3 2009 | | | Q2 2009 | | | Q1 2009 | |
| | (Dollars in millions) | |
Delinquent risk-in-force rescinded per quarter | | $ | 85.3 | | | $ | 78.7 | | | $ | 109.2 | | | $ | 123.6 | | | $ | 148.4 | | | $ | 170.3 | | | $ | 200.6 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| | Q3 2010 | | | Q2 2010 | | | Q1 2010 | | | Q4 2009 | | | Q3 2009 | | | Q2 2009 | | | Q1 2009 | |
| | (Dollars in millions) | |
Non-delinquent risk-in-force rescinded per quarter | | $ | 0.8 | | | $ | 0.3 | | | $ | 1.3 | | | $ | 4.8 | | | $ | 9.5 | | | $ | 66.3 | | | $ | 62.0 | |
1 | Please note that new proposed disclosure is italicized. |
22
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 23 of 33
The number of loans on which coverage has been rescinded by PMI increased in 2008 and 2009. Our inventory of files under review peaked in 2010 and, as a result, rescission levels are generally declining relative to prior periods.
3. When PMI rescinds coverage of a loan, we remove it from our calculation of PMI’s risk-in-force and insurance in force. In addition, if the rescinded loan was delinquent, we cease to include that loan in our default inventory and, therefore, do not incorporate that loan into our loss reserve estimates. Accordingly, past rescissions of delinquent loans have materially reduced our loss reserve estimates. In arriving at our loss reserve estimates, we also consider the effect of projected future rescission activity with respect to the existing inventory of delinquent insured loans. Projected future rescissions are materially reducing our current loss reserve estimates, although not to the same extent as in recent periods. In the second quarter of 2010, we reduced our estimate of future rescissions as a result of a decline in rescission activity beyond our expectations. To the extent future rescission activity is lower than projected, we would be required to increase loss reserves in future periods.
4. Upon receiving PMI’s notice of rescission with respect to a loan, the insured, either directly or through its servicer, may seek additional information as to our rescission and/or request reconsideration by challenging the bases of our decision to rescind coverage on specific loans (“loan-level challenge”). Although PMI’s policies do not contain provisions addressing reconsideration requests, we review all loan-level challenges we receive. In some cases, insureds’ loan-level challenges include new or additional information and/or considerations. If we decide to reverse a rescission after consideration of a loan-level challenge, we reinstate coverage of the loan after receipt of the applicable premium (“reinstatement”). If we reinstate coverage of a loan that is delinquent at the time of reinstatement, regardless of its status at the time of the rescission, we reassume the risk and include the loan in our delinquent loan inventory. If, as a result of our review, we affirm our initial rescission, we inform the customer of our decision. If levels of reinstatements exceed our expectations, our financial condition and results of operations will be negatively impacted. We believe that insureds typically do not notify PMI in the event that they consider their loan-level challenge to be resolved. In some cases, insureds are also challenging our general rights to rescind coverage of all or any loans under the terms of PMI’s master policies (“general challenges”). These general challenges do not identify the specific loans believed to be at issue and may refer to broad categories of rescissions.
5. If we are not able to informally resolve a disagreement with an insured regarding a rescission of coverage, the insured may seek resolution of the disagreement by filing a lawsuit or requesting that PMI agree to an arbitration. PMI’s mortgage insurance policies contain arbitration clauses providing for binding arbitration upon mutual consent of the parties. See Legal Proceedings for a discussion of certain rescission related litigation in which PMI is currently involved. The table below shows, for each year in
23
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 24 of 33
which the risk was written (“book year”), the aggregate delinquent and non-delinquent rescinded risk to date, net of reinstatements, and the aggregate risk amount of previously rescinded loans, net of reinstatements, which PMI believes may be subject to disagreement, either because PMI has received a loan-level challenge or because the loan is subject to pending litigation, arbitration or other dispute resolution process. As discussed above, because insureds generally do not notify PMI when they agree with our decisions to defend rescissions, an unknown portion of what we categorize as disputed rescinded risk may not actually be subject to disagreement.
Table 2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| | Primary and Pool* (by book year) | |
| | 2004 and Prior | | | 2005 | | | 2006 | | | 2007 | | | 2008 | | | 2009 | | | Total | |
| | (Dollars in millions) | |
Delinquent Risk: | | | | |
Rescinded Risk (Net of Reinstatements) | | $ | 130.5 | | | $ | 81.4 | | | $ | 343.6 | | | $ | 867.4 | | | $ | 100.5 | | | $ | 1.2 | | | $ | 1,524.6 | |
Rescinded Risk Disputed (Net of Reinstatements) | | | 11.2 | | | | 17.8 | | | | 68.1 | | | | 207.3 | | | | 25.6 | | | | 0.1 | | | | 330.1 | |
Rescinded Risk Disputed and Subsequently Re-Affirmed by PMI | | | 11.2 | | | | 15.2 | | | | 45.7 | | | | 128.4 | | | | 17.1 | | | | — | | | | 217.6 | |
Rescinded Risk Disputed and Currently Pending Investigation | | $ | 0.1 | | | $ | 2.6 | | | $ | 22.4 | | | $ | 78.9 | | | $ | 8.4 | | | $ | 0.1 | | | $ | 112.5 | |
Non-Delinquent Risk: | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
Rescinded Risk (Net of Reinstatements) | | $ | 31.6 | | | $ | 27.1 | | | $ | 67.8 | | | $ | 365.2 | | | $ | 6.5 | | | $ | 0.6 | | | $ | 498.8 | |
Rescinded Risk Disputed (Net of Reinstatements) | | | 2.2 | | | | 2.2 | | | | 14.3 | | | | 47.7 | | | | 1.4 | | | | — | | | | 67.8 | |
Rescinded Risk Disputed and Subsequently Re-Affirmed by PMI | | | 2.0 | | | | 2.2 | | | | 7.2 | | | | 16.0 | | | | 1.4 | | | | — | | | | 28.8 | |
Rescinded Risk Disputed and Currently Pending Investigation | | $ | 0.2 | | | $ | — | | | $ | 7.1 | | | $ | 31.6 | | | $ | 0.1 | | | $ | — | | | $ | 39.0 | |
* | Data excludes rescissions of pool risk subject to modified pool restructurings completed in 2009 and 2010. |
6. Because insureds’ general challenges to PMI’s rescission rights do not identify specific rescinded loans, the total population of rescinded loans that may be subject to general challenges is uncertain. Accordingly, we are unable to quantify the aggregate risk amount of rescinded loans that may be subject to general challenges. As we have received general challenges from several larger customers, the aggregate risk relating to general challenges would be material if we are unable to satisfactorily resolve such challenges. Moreover, because our or other mortgage insurers’ contractual rescission rights have been subject to few judicial decisions, it is unclear whether a court would adopt the same interpretation of our contractual rescission rights as we do. Accordingly, there is a risk that we will not be successful in defending our contractual bases of rescission against challenges. If this were to happen, we would be required to reassume rescinded risk and re-establish loss reserves on delinquent loans, which would negatively impact our financial condition and results of operations. See Risk Factors – Rescission activity and claim denials may not materially reduce our loss reserve estimates at the same levels we have recently experienced. Also, our loss reserves will increase if we reverse rescissions or claim denials beyond expected levels, if future rescission or claim denial activity is lower than projected, or if we are unable to
24
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 25 of 33
defend our rescissions and claim denials in litigation and, therefore, be required to reassume risk and establish loss reserves on delinquent loans.
7. | Note 2. Summary of Significant Accounting Policies |
Revenue Recognition – Mortgage guaranty insurance policies are contracts that are generally non-cancelable by the insurer, are renewable at a fixed price, and provide for payment of premiums on a monthly, annual or single basis. Upon renewal, the Company is not able to re-underwrite or re-price its policies. Consistent with industry accounting practices, premiums written on a monthly basis are earned as coverage is provided. Monthly premiums accounted for x.x%, 88.5% and 84.7% of gross premiums written from the Company’s mortgage insurance operations in 2010, 2009 and 2008, respectively. Premiums written on an annual basis are amortized on a monthly pro rata basis over the year of coverage. Primary mortgage insurance premiums written on policies covering more than one year are referred to as single premiums. A portion of the revenue from single premiums is recognized in premiums earned in the current period, and the remaining portion is deferred as unearned premiums and earned over the expected life of the policy, a range of seven to fifteen years for the majority of the single premium policies. If single premium policies related to insured loans are cancelled due to repayment by the borrower, and the premium is non-refundable, then the remaining unearned premium related to each cancelled policy is recognized as earned premiums upon notification of the cancellation. Unearned premiums represent the portion of premiums written that is applicable to the estimated unexpired risk of insured loans.A portion of premium payments may be refundable if the insured cancels coverage, which generally occurs when the loan is repaid, the loan amortizes to a sufficiently low amount to trigger a lender permitted or legally required cancellation, or the value of the property has increased sufficiently in accordance with the terms of the contract.In addition, when the Company pays a claim on a delinquent loan, all premiums received on the delinquent loan covering any period after the default date will be refunded, in accordance with the terms of the contract. In the event the Company rescinds insurance coverage of a loan as a result of an issue that occurred during the origination of the loan and/or prior to the effective date of coverage, it refunds all premiums associated with the rescinded loan to the insured, whether or not the loan was delinquent. Premium refunds reduce premiums earned in the Consolidated Statements of Operations.
25
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 26 of 33
Appendix B
Existing and Proposed Loss Reserve Disclosure
Management’s Discussion and Analysis of Financial Condition and Results of Operations
Critical Accounting Estimates
Reserves for Losses and LAE
1. The following table shows the reasonable range of losses and LAE reserves, as determined by our actuaries, and recorded for losses and LAE (gross of recoverables) as of September 30, 2010 and December 31, 2009.
| | | | | | | | | | | | | | | | | | | | | | | | |
| | As of September 30, 2010 | | | As of December 31, 2009 | |
| | Low | | | High | | | Recorded | | | Low | | | High | | | Recorded | |
| | (Dollars in millions) | | | (Dollars in millions) | |
U.S. Mortgage Insurance Operations | | $ | 2,600.0 | | | $ | 3,550.0 | | | $ | 2,982.5 | | | $ | 2,650.0 | | | $ | 3,850.0 | | | $ | 3,213.7 | |
International Operations | | | 25.5 | | | | 43.6 | | | | 32.7 | | | | 26.3 | | | | 46.1 | | | | 40.1 | |
| | | | | | | | | | | | | | | | | | | | | | | | |
Consolidated loss and LAE reserves* | | $ | 2,625.5 | | | $ | 3,593.6 | | | $ | 3,015.2 | | | $ | 2,676.3 | | | $ | 3,896.1 | | | $ | 3,253.8 | |
| | | | | | | | | | | | | | | | | | | | | | | | |
* | May not total due to rounding. |
2.U.S. Mortgage Insurance Operations —We establish PMI’s reserves for losses and LAE based upon our estimate of unpaid losses and LAE on (i) reported mortgage loans in default and (ii) estimated defaults incurred but not reported (“IBNR”) to PMI by its customers. Our best estimate of PMI’s reserves for losses and LAE is derived primarily from our analysis of PMI’s default and loss experience. The key assumptions used in the estimation process are expected claim/cure rates, average claim sizes and costs to settle claims. We evaluate our assumptions in light of PMI’s historical patterns of claim payments, loss experience in past and current economic environments, the seasoning of PMI’s various books of business, PMI’s coverage levels, the credit quality profile of PMI’s portfolios, and the geographic mix of PMI’s business. Our assumptions are influenced by historical loss patterns and are adjusted to reflect recent loss trends. Our assumptions are also influenced by our assessment of current and future economic conditions, including trends in housing prices and unemployment. We also evaluate various scenarios representing possible losses and LAE under different economic assumptions.
3. The Company’s actuaries utilize a number of generally recognized actuarial projection methodologies in order to determine a single point estimate of loss reserves. The actuarial reviews and documentation are completed in accordance with professional actuarial standards with reserves established on a basis consistent with U.S. GAAP. The selected assumptions reflect the actuary’s judgment based on historical data and experience combined with information concerning current underwriting, economic, judicial, regulatory and other influences on ultimate claim settlements.
4. The lead methodology our actuaries use is a frequency-severity method based on the inventory of pending delinquencies. The frequency-severity method is preferable when there
26
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 27 of 33
have been significant changes in established historical patterns of losses, as is the case in the recent and current credit cycle. In using this method, our actuaries can optimize results by utilizing our knowledge of the number of delinquent loans and the coverage amounts (“risk size”) on those loans. The delinquencies are grouped by the year in which they were reported to PMI (“report year”). A claim rate is then developed for each report year, which represents an estimate of the frequency with which the delinquencies may become claims. The claim rates are based on an analysis of the patterns of emerging cure counts and claim counts for the report year, the foreclosure status of the pending delinquencies, the product and geographical mix of the delinquencies and the actuaries’ views of future economic and claim conditions. The actuaries estimate the severity of projected claims by examining the risk sizes on the delinquent loans and estimating the portion of the coverage that will be paid, as well as any expenses. The actuaries’ estimation of the portion of coverage paid is based on a review of historical data and an assessment of economic conditions and consideration of the opportunities for loss mitigation techniques, various expenses claimable by insureds and the level of equity the borrowers may have in their homes.
5. In addition to the frequency-severity method, our actuaries use various chain-ladder methods utilizing actual loss amounts and expected patterns of loss emergence to determine an estimate of ultimate losses. These methods may be appropriate when there is a relatively stable pattern of loss emergence. Chain-ladder techniques are less suitable in cases in which the insurer does not have a developed claims history. In the chain-ladder technique, the ultimate value of a claim count, paid loss amount or other amount is estimated by applying a loss development factor to the current value. The loss development factor measures the portion of the loss emergence pattern that has been completed and extends the current amount to its ultimate value. The difference between the ultimate value and the current value is the amount carried in the loss reserve. The emergence pattern used in this technique is based on an analysis of historical patterns of emergence from prior report years.
6. Management believes the amounts recorded represent the most likely outcome within the actuarial ranges. The recorded reserves for U.S. Mortgage Insurance Operations as of September 30, 2010 are below the midpoint of the actuarially-determined reserve range. Management’s best estimate was below the actuarial midpoint primarily due to its views with respect to the ultimate cure rates of our pending delinquencies. The resolution of a significant number of delinquencies in our insured loan inventories has been delayed beyond historical levels. Management believes that the likely ultimate benefit of the resolution of these delinquencies, including through retention loan work-outs, including loan modifications, and borrowers bringing their loans current, is not fully reflected in available data and, accordingly, was not fully considered in the claim rates used in the development of the actuarial ranges.
7. When our actuaries develop an actuarial range of losses and LAE and their estimates of claim rates, they place significant reliance on historical data on completed loan modifications and other loan work-outs. In management’s estimate of claim rates, management evaluates the same historical data, but also considers contemporaneous factors occurring in the broader mortgage market that may not yet be reflected in the historical data, including
27
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 28 of 33
specific pending loan modifications and other retention loan work-outs reported by servicers and more recent observations of resolution of delinquent loan trends. Accordingly, management attempts to identify trends which may not be fully present in data sets used to project actuarial loss estimates. To the extent future actual claim rates are above management’s expectations, we would expect that future reserves would need to increase from current levels. In the third quarter of 2010, we increased our future expected claim rates primarily as a result of lower than expected cures in 2010, which negatively impacted loss reserves in the period. There is no assurance that we will not have to adjust claim rates in the future based on actual cures on our existing delinquent loan inventories.
8. Our projection of future modifications of delinquent insured loans is a factor in management’s evaluation of the selected claim rate. We continue to expect that a material percentage of loans in our delinquent loan inventory will be modified or become current, and management’s loss reserve estimation process takes into consideration this expectation. Because the age of a delinquency is a primary factor in determining whether a loan is likely to modify and become current in the future, management develops its projections of future modifications of existing delinquent loans based on the year in which we received the notice of default (an “NOD report year”). Within each NOD report year, management divides the pending delinquent loans into various categories depending on their relative probabilities of modifying and curing in the future. To develop the categories and probabilities, management assesses numerous factors, including servicer reports on the stages of retention and liquidation workouts and forbearance arrangements; management’s assessment of loans in our default inventory that may have the potential to qualify for an available modification plan, including HAMP; and historical modification data and trends. As of December 31, 2010, we forecast that approximately xx % of our primary delinquent loan inventory will cure as a result of future loan modifications. The estimated amount of risk in force associated with this percentage was approximately $x.x billion.
9.Rescission activity and loss reserves. Our projection of future rescission activity with respect to the current inventory of delinquent loans has reduced and continues to materially reduce our loss reserve estimates. In each reporting period, management estimates the future rescission rate on the then existing inventory of delinquent loans. Management’s estimate is based upon historical data and trends, including past percentages of delinquent loans rescinded, and other factors, including the characteristics and age of the delinquent loan inventory and the current inventory of loan files under or scheduled for investigation. In this estimation process, we also consider that we will in the future reinstate a portion of the policies subject to future rescission activity following requests by lenders for reconsideration or challenges of our rescission decisions. (For a discussion of our rescission reconsideration process, seeManagement’s Discussion and Analysis of Financial Condition and Results of Operations – Conditions and Trends Affecting our Business – U.S.M.I. – Rescission Activity.) Accordingly, in
28
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 29 of 33
our forecasting, we utilize an estimate of future rescissions net of estimated reinstatements. Our practice of projecting reinstatements has been refined over time as we gain further actual experience and utilizes, among other things, our experience of the cumulative incidence of rescission reinstatements that have also become paid claims and our views of likely future trends. As of December 31, 2010, our loss reserves balance was reduced by approximately $x.x billion as a result of our estimated future rescission activity (net of estimated reinstatements).
10. As a result of our rescission reconsideration process, management also expects that PMI will reinstate coverage in the future on a portion of previously rescinded policies subject to requests by lenders that we reconsider our rescission. At any point in time, we have pending requests or challenges from lenders with respect to which we have not completed our reconsideration of our rescission decision. Based on our historical experience, we estimate the portion of policies that we expect to reinstate after our reconsideration is complete, and take such estimate into account in establishing our IBNR reserves. With respect to loans as to which we have determined not to reinstate coverage, we do not include a reserve in our IBNR and loss reserves for the possibility that we may be unsuccessful in defending our rescissions in litigation or other dispute resolution processes.
11. Our investigation and rescission reconsideration processes require detailed reviews of loan level origination files, which are time and resource intensive. As a result, during any reporting period, we maintain inventories of: (i) loan files identified for investigation or files which are being investigated; and (ii) requests for PMI to reconsider its previous rescissions of coverage which are under review by PMI. We review the size and characteristics of these inventories in detail when we project future rescission and rescission reinstatement activity. In 2010, these inventories declined, thereby reducing our projected future rescission and rescission reinstatement activity. SeeRisk Factors – Rescission activity and claim denials may not materially reduce our loss reserve estimates at the same levels we have recently experienced. Also, our loss reserves will increase if we reverse rescissions or claim denials beyond expected levels, if future rescission or claim denial activity is lower than projected, or if we are unable to defend our rescissions or claim denials in litigation and, therefore, be required to reassume risk and establish loss reserves on delinquent loans.
12. Claim denials and loss reserves.Our projection of future claim denial activity with respect to the current inventory of delinquent loans has reduced, and is continuing to reduce, our loss reserve estimates. Historically, the significant majority of claim denials resulted from the inability of servicers to produce documents necessary to perfect the claim (“documentation claim denials”). Management’s estimate of future documentation claim denials on the then existing inventory of delinquent loans is based upon historical document claim denial data and trends, and the performance of various servicers of loans within the delinquent loan inventory. In arriving at its estimate, management also estimates the percentage of future document claim
29
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 30 of 33
denials that will later be reversed, and therefore paid, as a result of the servicer ultimately providing to PMI the required loan or claim documentation. In considering future reversal rates, management reviews historical reversal data by particular servicers.
13. Our estimate of future reversals of claims previously denied is one of the estimates that ultimately result in the setting of PMI’s IBNR reserve. For PMI’s IBNR reserve as of December 31, 2010, the methodology we used to estimate future reversals of claims previously denied was similar to our rescission reversal estimate process described above.
14. Beginning in 2010, denials also include claims denied, pended or curtailed as a result of the negligent servicing of delinquent loans by servicers (“servicer-related claim denials”). We expect the number of servicer-related claim denials to significantly increase in 2011. Beginning in the fourth quarter of 2010, we began to estimate within our loss reserving process the effect of future servicer-related claim denials on our existing inventory of delinquent loans. Our estimate was based upon our internal reviews of various servicers’ performance as well as samples of delinquent loans. As with documentation claim denials, our estimate includes assumptions relating to the percentage of servicer-related denials that will later be reversed and paid. In considering future reversal rates, we reviewed, among other things, historical reversal data related to documentation claim denials.
15. As of December 31, 2010, our loss reserve balance was reduced by approximately $x.x billion by our estimated future claim denial activity (net of estimated reversals of claim denials). SeeRisk Factors – Rescission activity and claim denials may not materially reduce our loss reserve estimates at the same levels we have recently experienced. Also, our loss reserves will increase if we reverse rescissions or claim denials beyond expected levels, if future rescission or claim denial activity is lower than projected, or if we are unable to defend our rescissions or claim denials in litigation and, therefore, be required to reassume risk and establish loss reserves on delinquent loans.
30
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 31 of 33
16. The table below provides a reconciliation of our U.S. Mortgage Insurance Operations’ beginning and ending reserves for losses and LAE for the years ended December 31, 2009, 2008 and 2007:
| | | | | | | | | | | | |
| | 2009 | | | 2008 | | | 2007 | |
| | (Dollars in millions) | |
Balance at January 1 | | $ | 2,624.5 | | | $ | 1,133.1 | | | $ | 366.2 | |
Reinsurance recoverables | | | (482.7 | ) | | | (35.9 | ) | | | (2.9 | ) |
| | | | | | | | | | | | |
Net balance at January 1 | | | 2,141.8 | | | | 1,097.2 | | | | 363.3 | |
Losses and LAE incurred (principally with respect to defaults occurring in): | | | | | | | | | | | | |
Current year | | | 1,552.7 | | | | 1,688.0 | | | | 894.4 | |
Prior years | | | 187.9 | | | | 147.7 | | | | 201.7 | |
| | | | | | | | | | | | |
Total incurred | | | 1,740.6 | | | | 1,835.7 | | | | 1,096.1 | |
Losses and LAE payments (principally with respect to defaults occurring in): | | | | | | | | | | | | |
Current year | | | (308.7 | ) | | | (100.9 | ) | | | (43.6 | ) |
Prior years | | | (1,063.6 | ) | | | (713.1 | ) | | | (318.6 | ) |
| | | | | | | | | | | | |
Total payments | | | (1,372.3 | ) | | | (814.0 | ) | | | (362.2 | ) |
| | | | | | | | | | | | |
Transfer of PMI Guaranty loss reserves to U.S. Mortgage Insurance | | | — | | | | 22.9 | | | | — | |
Net balance at December 31 | | | 2,510.1 | | | | 2,141.8 | | | | 1,097.2 | |
Reinsurance recoverables | | | 703.6 | | | | 482.7 | | | | 35.9 | |
| | | | | | | | | | | | |
Balance at December 31 | | $ | 3,213.7 | | | $ | 2,624.5 | | | $ | 1,133.1 | |
| | | | | | | | | | | | |
17. The above loss reserve reconciliation shows the components of our losses and LAE reserve changes for the periods presented. Losses and LAE payments of $1.4 billion, $814 million and $362.2 million for the years ended December 31, 2009, 2008 and 2007, respectively, reflect amounts paid during the periods presented and are not subject to estimation. Total losses and LAE incurred of $1.7 billion, $1.8 billion and $1.1 billion for the years ended December 31, 2009, 2008 and 2007, respectively, are management’s best estimates of ultimate losses and LAE on our existing delinquent loan inventory and are subject to change. The $1.6 billion of total losses and LAE incurred with respect to defaults occurring in the current year for the year ended December 31, 2009 was approximately $0.1 billion higher due to an increase in estimated claim rates for those respective reported defaults. The changes in our estimates are principally reflected in the “Losses and LAE incurred (principally with respect to defaults occurring in)” line item which shows an increase to losses and LAE incurred related to prior years of $187.9 million, $147.7 million and $201.7 million for the years ended December 31, 2009, 2008 and 2007, respectively. PMI’s reserve balance at January 1, 2009 includes $22.9 million of PMI Guaranty’s loss reserves, which were merged into our U.S. Mortgage Insurance Operations in 2008.
31
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 32 of 33
18. The table below provides the changes in reserves related to prior years by particular accident years for 2009, 2008 and 2007, respectively:
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| | Losses and LAE Incurred | | | Change in Incurred | |
Accident Year (year in which default occurred) | | 2009 | | | 2008 | | | 2007 | | | 2006 | | | 2009 vs. 2008 | | | 2008 vs. 2007 | | | 2007 vs. 2006 | |
| | (Dollars in millions) | | | | |
2002 and prior | | $ | — | | | $ | — | | | $ | — | | | $ | — | | | $ | (8.6 | ) | | $ | (0.6 | ) | | $ | 3.1 | |
2003 | | | 226.5 | | | | 225.9 | | | | 226.8 | | | | 220.2 | | | | 0.6 | | | | (0.9 | ) | | | 6.6 | |
2004 | | | 241.1 | | | | 241.2 | | | | 241.3 | | | | 229.1 | | | | (0.1 | ) | | | (0.1 | ) | | | 12.2 | |
2005 | | | 271.9 | | | | 272.0 | | | | 270.8 | | | | 240.4 | | | | (0.1 | ) | | | 1.2 | | | | 30.4 | |
2006 | | | 414.3 | | | | 409.9 | | | | 410.3 | | | | 260.9 | | | | 4.4 | | | | (0.4 | ) | | | 149.4 | |
2007 | | | 1,114.7 | | | | 1,041.6 | | | | 893.1 | | | | — | | | | 73.1 | | | | 148.5 | | | | — | |
2008 | | | 1,829.5 | | | | 1,710.9 | | | | — | | | | — | | | | 118.6 | | | | — | | | | — | |
2009 | | | 1,552.8 | | | | — | | | | — | | | | — | | | | — | | | | — | | | | — | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
Total | | | | | | | | | | | | | | | | | | $ | 187.9 | | | $ | 147.7 | | | $ | 201.7 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| | | | | | | |
| | | | | | | | | | | | | | 2009 | | | 2008 | | | 2007 | |
| | | | | | | | | | | | | | | | | | | (Dollars in millions) | |
Primary insurance portfolio: | | | | | | | | | | | | | | | | | | $ | (72.5 | ) | | $ | 111.5 | | | $ | 189.2 | |
Pool insurance portfolio: | | | | | | | | | | | | | | | | | | | 260.4 | | | | 36.2 | | | | 12.5 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
Changes in reserves related to prior years | | | | | | | | | | | | | | | | | | $ | 187.9 | | | $ | 147.7 | | | $ | 201.7 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
The $187.9 million, $147.7 million and $201.7 million increases related to prior years’ reserves in 2009, 2008 and 2007, respectively, were due to re-estimations of ultimate loss rates from those established at the original notices of defaults, updated through the periods presented. The $187.9 million increase in prior years’ reserves during 2009 reflected the significant weakening of the housing and mortgage markets and was driven primarily by adverse claim size and claim rate development in the modified pool insurance portfolio which resulted in $260.4 million of adverse development partially offset by $72.5 million from our primary insurance portfolio related to reductions in claim size on our primary insurance portfolio. The adverse development experienced in 2009 with respect to pool losses would have been approximately $130 million higher if not for the positive impact on loss reserves related to negotiated early discounted claim payments for modified pool contract restructurings completed in 2009. The $147.7 million increase in prior years’ reserves during 2008 and the $201.7 million increase in prior years’ reserves in 2007 reflected significant weakening of the housing and mortgage markets in 2008 and 2007 and were driven by lower cure rates, higher claim rates and higher claim sizes related to our primary insurance portfolio in those years. These re-estimations of ultimate loss rates are
32
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
January 31, 2011
Page 33 of 33
the result of management’s periodic review of estimated claim amounts in light of actual claim amounts, loss development data and ultimate claim rates. Future declines in PMI’s cure rate, or higher default and claim rates or claim sizes could lead to further increases in losses and LAE.
33