Triad Guaranty Inc.
101 South Stratford Road
Winston-Salem, NC 27104
June 27, 2007
Mr. Jim B. Rosenberg
Senior Assistant Chief Accountant
United States Securities and Exchange Commission
Division of Corporation Finance
Mail Stop 03-09
450 Fifth Street, N.W.
Washington, DC 20549
Re: | Triad Guaranty Inc. |
| Form 10-K for the Year Ended December 31, 2005 |
| File No. 000-22342 |
Dear Mr. Rosenberg:
We respectfully submit the following in response to our telephone conversation on June 13, 2007 with Ms. Kei Ino in which she sought further clarification on information contained in our responses dated January 31, 2007, March 23, 2007 and May 4, 2007. Ms. Ino’s comments are numbered below and our response is shown immediately below each numbered comment in boldface type:
| 1. | In disclosure format, discuss the types of mortgages giving rise to the reported deficiencies during the years presented. Also please discuss why the reserve calculation methodology was not changed to incorporate a separate segment for pending claims until 2006, even though pending claims began to increase in 2004. |
As disclosed in the Annual Report on Form 10-K, we segment our defaults into groups with common characteristics, where such characteristics are believed to be meaningful predictors for estimating the number and severity of the claims that ultimately will be paid. Historically the type of mortgage (fixed or adjustable rate) has not been a differentiating factor in our reserve model once a loan is in default. The significant majority of the inventory of loans in default for the 2004 – 2006 period were fixed rate or amortizing adjustable rate loans that result in similar principle reduction amounts over time.
As we have mentioned in previous correspondence, prior to 2006 our reserve calculation did not include a separate segment for pending claims. Our
1
frequency factors incorporated the estimated mix of pending claims into the overall frequency factor for each segment. We were aware that, when a significant change in the mix towards more pending claims occurred, the historical reserve factors had the potential to understate reserves. Therefore prior to 2006, in calculating the estimate of the reserve for losses, consideration was given to the level of pending claims, even though they were not included as a separate segment. Historically, this judgmental process was considered to be appropriate due to the relative stability of the level of pending claims within the inventory of defaults. However, because of the volatility experienced in the level of pending claims in recent periods, during 2006 we changed our reserve estimation process to treat pending claims as a separate segment with its own frequency factor. We believe the addition of a pending claim segment provides a better estimate of a key event in the loss recognition cycle.
| 2. | In your previous responses, you indicated that you changed the estimates for frequency and severity utilized in the reserve methodology during 2005, and these changes had the impact of increasing the reported deficiency for December 31, 2004 defaults that remained in default at December 31, 2005. Please explain why it was appropriate to apply trends that arose in 2005 to defaults existing as of December 31, 2004. |
The estimate of the reserve for losses is intended to cover the full cost of settling claims on loans in default. Changes in economic conditions that occur after the financial reporting date, but prior to the settlement of the claims, may impact the ultimate cost of settling claims on loans in defaults, requiring a change in the estimate of the reserve for losses, regardless of the date of default. Changes in economic conditions that we believe will influence the ultimate cost of settling claims are incorporated in our estimate of the reserve for losses and reflected in our financial statements as such changes occur, in accordance with generally accepted accounting principles.
| 3. | In your response dated May 4, 2007, you provided a schedule of the weighted average frequency and severity factors incorporated in the calculation of the estimate of the reserve for losses at December 31, 2006, 2005 and 2004. Based on this schedule, at December 31, 2004, the average severity was $20,055 and the average frequency was 32.2%. In your Annual Report on Form 10-K for the year ended December 31, 2006, you disclose on page 44 the number of loans in default at December 31, 2004 totaled 5,922. Multiplying the loans in default at December 31, 2004 by the average frequency and severity factors in your May 4, 2007 response produces a reserve of $38.2 million. Please explain why this calculation produces reserves significantly higher than the $34.0 million reported at December 31, 2004. |
At December 31, 2004, the total number of defaults was 5,922, including 634 defaulted loans that were part of structured bulk transactions subject to deductibles. As disclosed on page 45 of our Management’s Discussion and Analysis section of the Annual Report on Form 10-K for the year ended December 31, 2006, we do not provide reserves on loans that are part of bulk
2
transactions structured with deductibles until the incurred losses for any specific transaction are expected to exceed the deductibles. Also as disclosed on page 45, no individual structured bulk transaction with a deductible as part of the structure had incurred losses that were projected to exceed the deductible and, excluding the 634 defaulted loans subject to deductible structures, results in 5,288 reserved defaults. Using the methodology suggested in your question, and applying the overall average frequency and average severity factors to the 5,288 reserved defaults, would result in an amount of $34.1 million, as illustrated below. This compares to the actual reserve for reported defaults of $30.3 million. The $30.3 million is the reported case reserve at December 31, 2004 while the $34.0 million you reference above is the total reserve, including reserves for incurred but not reported defaults and loss adjustment expenses.
Reserved defaults | 5,288 |
Frequency factor | 32.2% |
Average severity | $20,055 |
Projected case reserve for reported defaults (in 000’s) | $34,148 |
Actual case reserve for reported defaults at December 31, 2004 (in 000’s)* |
$30,271 |
* Reserve excluding provision for defaults incurred but not reported and loss adjustment expenses. Applying the frequency factor and the severity factor to the reported reserved defaults produces an estimate of our actual case reserve only.
The variance in the projected case reserve utilizing the methodology suggested in your question and the actual case reserve for reported defaults is due to the method used in our response of May 4, 2007 to calculate the weighted average frequency and severity factors. In our response, we utilized the ending reserve balance to calculate the weighted average factors, as opposed to the default counts. Due to the differences in the characteristics of the business insured in the primary and modified pool lines, with a principal difference being significantly lower loan to value ratios in the modified pool line, the frequency and severity factors historically had been lower for the modified pool lines. If we had used default counts by line as the weighting factor in our response of May 4, 2007, the weighted average frequency and severity factors are 29.6% and $19,316, respectively. Applying those factors to the 5,288 reserved defaults produces a reserve of $30.2 equal to the actual reserve for reported defaults shown in the table above. In calculating the weighted average severity and frequency factors that we utilized in our response of May 4, 2007, we believe it was more accurate to apply a weighting factor utilizing reserves rather than the count.
| 4. | As reported in the statistical information included with your earnings news release dated January 27, 2005, the average severity for paid claims for the quarter ended December 31, 2004 was disclosed as $24,100 for the flow business line and $27,000 for the bulk business line. Please explain and/or reconcile the severity experienced in the fourth |
3
quarter of 2004 to the overall average severity of $20,055 incorporated in the December 31, 2004 calculation of the estimate of the reserve for losses.
Average Paid Severity by Quarter for 2004
(000’s) | 1st Qtr 2004 | 2nd Qtr. 2004 | 3rd Qtr. 2004 | 4th Qtr 2004 |
Flow | $27.4 | $25.6 | $21.2 | $24.1 |
Bulk | $30.2 | $22.5 | $17.6 | $27.0 |
Total | $27.7 | $25.1 | $20.6 | $24.8 |
The severity assumption utilized in the calculation of the estimate of reserve for losses reflects recent trends in claims payments and mitigation results. The economic conditions in 2004 were largely favorable for the housing and mortgage insurance industry, with those favorable conditions being reflected in our improving severity. At the time of the December 31, 2004 calculation of the estimate for the reserve for losses, the recent trends had been toward improving severity, with the exception of the fourth quarter of 2004. Additionally in the 2004 time frame the volume of claims paid on a quarterly basis was smaller, and more subject to being impacted by a few large claims. The impact of large claims was particularly meaningful to the bulk line due to the low volume of paid claims in that line, with only 202 claims paid in the bulk line during 2004. Therefore, in analyzing the actual severity we considered the impact of the large claims on a given quarter. In the fourth quarter of 2004 there was a significant increase in large claims paid (at the time claims larger than $45,000 were considered large claims) in the bulk line of business that influenced bulk and overall severity. For the first three quarters of 2004 we averaged five large paid claims per quarter in the bulk line; however, in the fourth quarter of 2004 eighteen large claims were paid. Excluding large claims our severity trends were:
| Average Paid Severity Excluding Large Claims |
(000’s) | 1st Qtr 2004 | 2nd Qtr. 2004 | 3rd Qtr. 2004 | 4th Qtr 2004 |
Flow | $22.1 | $21.2 | $17.3 | $19.6 |
Bulk | $21.5 | $16.7 | $15.8 | $19.2 |
Total | $22.0 | $20.6 | $17.0 | $19.5 |
Large claims paid in the fourth quarter of 2004 significantly increased the bulk and overall severity, but did not change the general trends in severity during 2004. The development of the estimate of reserve for losses requires judgmental interpretations of trends, including when a change in data point may mark a change in the trend. Given the significant improvement in severity observed prior to the fourth quarter of 2004, at the time of the reserve calculation in early 2005, we believed the general trends observed through 2004 were more
4
representative of the expected severity than the severity experienced in the fourth quarter of 2004. With the benefit of hindsight it is clear that the fourth quarter of 2004 was an inflection point in terms of severity trends; however, at the time of the calculation of the estimate of the reserve for losses our assumption, using the information we had available at that time, was that severity trends would be similar to that experienced in the first through third quarter of 2004.
While there are many details that can be discussed to explain the deficiency in the estimate of the reserve for losses, with hindsight we now believe an underlying reason was a change in the sources of our business. During the 2001 - 2002 period we began to add business from new and different sources, including large mortgage lenders which we previously had little business with, and structured bulk transactions. As this business aged and seasoned our default inventory began to reflect defaults from the new sources of business, creating concentrations in our defaults that did not previously exist. For example, at December 31, 2001, 85% of our default inventory was from our traditional flow business line (which consisted of many, relatively small lenders) and by December 31, 2004, the volume of defaults from our traditional business line had declined to 37% of reserved defaults, with the balance being comprised of defaults from two lenders (38% of reserved defaults) and the structured bulk business (25% of reserved defaults). In retrospect, the business from the new sources had fundamental differences in quality that impacted performance that we initially did not fully incorporate in our estimate of the reserve for losses. In developing the factors underlying the estimate of the reserve for losses we used our historical experience and other performance variables, such as the losses assumed in our pricing of the business, to develop our expectations for frequency and severity. With the change in the composition of defaults, the reserve factors used to calculate the estimate of the reserve for losses proved not fully representative of the actual loss experience on the December 31, 2004, inventory of defaults, contributing to the deficiency in reserves that developed in the ensuing years.
5
We confirm and acknowledge that:
| § | we are responsible for the adequacy and accuracy of the disclosure in our filing; |
| § | staff comments or changes to disclosure in response to staff comments do not foreclose the Commission from taking any action with respect to the filing; and |
| § | we 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. |
If you have any questions concerning our response, please do not hesitate to contact me at (336) 723-1282, extension 1105 or contact Earl Wall, our general counsel, at extension 1108.
Sincerely,
/s/ Kenneth W. Jones
Kenneth W. Jones
Senior Vice President and Chief Financial Officer
6
