Losses and LAE

Our reserve for losses and LAE, as of the dates indicated, consisted of:



(In thousands)	September 30, 2013		December 31, 2012
Mortgage insurance reserves	$	2,314,785	$	3,083,608
Financial guaranty reserves	32,094		66,328
Total reserve for losses and LAE	$	2,346,879	$	3,149,936

The following table presents information relating to our mortgage insurance reserves for losses, including our estimate of defaults incurred but not reported (“IBNR”), and LAE as of the dates indicated:



	Three Months Ended September 30,					Nine Months Ended September 30,
(In thousands)	2013		2012			2013			2012
Mortgage Insurance
Balance at beginning of period	$	2,690,861	$	3,155,343		$	3,083,608		$	3,247,900
Less reinsurance recoverables (1)	58,427		97,845			83,238			151,569
Balance at beginning of period, net of reinsurance recoverables	2,632,434		3,057,498			3,000,370			3,096,331
Add losses and LAE incurred in respect of default notices reported and unreported in:
Current year (2)	140,015		248,806			477,155			686,080
Prior years	11,997		(77,001		)	(56,777		)	(71,468		)
Total incurred	152,012		171,805			420,378			614,612
Deduct paid claims and LAE related to:
Current year (2)	21,334		2,912			21,483			3,185
Prior years	498,002		269,486			1,134,155			750,853
Total paid	519,336		272,398			1,155,638			754,038
Balance at end of period, net of reinsurance recoverables	2,265,110		2,956,905			2,265,110			2,956,905
Add reinsurance recoverables (1)	49,675		89,801			49,675			89,801
Balance at end of period	$	2,314,785	$	3,046,706		$	2,314,785		$	3,046,706

_________________________


(1)	Related to ceded losses on captive reinsurance transactions, capital markets reinsurance transactions (“Smart Home”) and QSR transactions.


(2)	Related to underlying defaulted loans with a most recent date of default notice in the year indicated. For example, if a loan had defaulted in a prior year, but then subsequently cured and later re-defaulted in the current year, that default would be considered a current year default.

The ultimate amount and timing of future losses will depend, in part, on general economic conditions and other factors, including the status of credit markets, home prices and unemployment rates, all of which are difficult to predict and beyond our control. Our mortgage insurance incurred losses are driven primarily by new mortgage insurance defaults and any changes in the assumptions used to determine our loss reserves on new or existing defaults.

Our mortgage insurance loss reserves declined in the third quarter of 2013 primarily due to a decrease in our total inventory of defaults as a result of the Freddie Mac Agreement. Additionally, the volume of paid claims, defaulted loans that cured (“cures”), and net insurance rescissions and claim denials outpaced new default notices received during the quarter, which also contributed to the decrease in defaults. Total paid claims increased for the three and nine months ended September 30, 2013 from the comparable periods in 2012, mainly as a result of the claim payment associated with the Freddie Mac Agreement and an improvement in the efficiency of our claim review process. In prior periods, our process for reviewing claims received for non-compliance with our insurance policies lengthened the claim resolution period. Delays created by foreclosure slowdowns, servicer issues, and loan modification programs also had lengthened the claim resolution process in prior periods. We cannot be certain of the ultimate impact of these delays on our business or results of operations.

In August 2013, Radian Guaranty entered into the Freddie Mac Agreement related to a group of 25,760 first-lien mortgage loans guaranteed by Freddie Mac that were insured by Radian Guaranty and were in default as of December 31, 2011. The Freddie Mac Agreement provides for the future treatment of the loans subject to the terms of the agreement including claim payments, loss mitigation activity and insurance coverage, and eliminates Radian Guaranty’s claim exposure on 9,756 loans that were delinquent and 4,586 loans that were re-performing as of July 31, 2013. The remaining loans in the original population as of December 31, 2011 were paid off, rescinded, denied or resulted in a paid claim prior to July 31, 2013. The Freddie Mac Agreement caps Radian Guaranty’s total exposure on the entire population of loans subject to the agreement to $840 million, and Radian Guaranty has no additional exposure to claims on these loans. The maximum exposure of $840 million is comprised of $625 million of claim payments (consisting of $370 million of claims paid on this population as of July 12, 2013 and $255 million paid at closing) and $215 million related to Loss Mitigation Activity on the loans. At the closing, Radian Guaranty deposited $205 million into a collateral account to cover future Loss Mitigation Activity on the loans. The amount deposited in the collateral account represents $215 million, less $10 million of Loss Mitigation Activity that had become final in accordance with the Freddie Mac Agreement prior to the date the collateral account was established. The collateral account consists of investment securities and remains on our consolidated balance sheets as a result of the rights that Radian Guaranty has with respect to those funds, as further described below. Subject to certain conditions in the Freddie Mac Agreement, amounts in the collateral account will be released to Radian Guaranty over time to the extent that Loss Mitigation Activity becomes final in accordance with the terms of the Freddie Mac Agreement. From the time the collateral account was established through September 30, 2013, approximately $2.4 million of additional Loss Mitigation Activity had become final in accordance with the Freddie Mac Agreement and $137.3 million of submitted claims had been rescinded, denied, curtailed or cancelled, but were not considered final in accordance with the Freddie Mac Agreement. If the amount of Loss Mitigation Activity that becomes final in accordance with the Freddie Mac Agreement after the collateral account was established does not accumulate to $205 million prior to termination of the Freddie Mac Agreement, then any remaining funds will be paid to Freddie Mac. Radian Guaranty will continue to administer all claims submitted with respect to these loans in accordance with the applicable insurance policy for these loans and in a manner consistent with its normal claims handling practices. The Freddie Mac Agreement will terminate upon the earliest to occur of: (1) August 29, 2017; (2) any time after August 29, 2015 if the amounts remaining in the collateral account are reduced to $0; or (3) any time after August 29, 2015 if Radian Guaranty exercises its early termination option to conclude the transactions under the Freddie Mac Agreement, by paying to Freddie Mac an amount equal to the initial collateral amount less the amount of Loss Mitigation Activity that had then become final under the terms of the agreement.

For the three and nine months ended September 30, 2013, reserves established for new default notices reported in the current year were the primary driver of our incurred losses. The adverse development in prior year incurred losses for the three and nine months ended September 30, 2013 includes the approximate $22 million initial loss related to the Freddie Mac Agreement. This initial loss is expected to be fully offset by a reduction of incurred losses in future periods. This future reduction of incurred losses is expected to result from the elimination of exposure to re-performing loans covered by the transaction that we expect to re-default in the future and ultimately become claims. Favorable reserve development on other default notices reported in prior years partially mitigated the impact from new defaults and the Freddie Mac Agreement, as the benefit to prior year defaults from higher claim curtailments and cures was more than previously estimated. In particular, claim curtailments due to servicer noncompliance with our insurance policies and servicing guidelines increased to approximately $16 million and $42 million, respectively, for the three and nine months ended September 30, 2013, compared to approximately $6 million and $15 million, respectively, for the three and nine months ended September 30, 2012.

For the three and nine months ended September 30, 2012, reserves established for new default notices were the primary driver of total incurred losses. Partially offsetting those effects in 2012 was positive reserve development on default notices reported in prior years, primarily relating to higher actual insurance rescissions and claim denials than previously assumed in our loss reserve estimates.

In addition, our results for the nine months ended September 30, 2012 were impacted by a $43.6 million decrease in our estimated reinsurance recoverable from our Smart Home transactions. This decrease was a result of trends in lower claims paid and higher insurance rescissions and claim denials than were previously estimated to occur by the scheduled maturity dates of our Smart Home transactions.

In recent years, our primary defaulted inventory experienced an increase in its weighted average age, as measured by the number of monthly payments missed. Because we apply higher estimated “default to claim rates” (rate at which defaulted loans are expected to result in claim) on our more aged delinquent loans, this has resulted in a higher reserve per default. In addition, our estimated rates of insurance rescissions and claim denials have declined in recent periods. Our aggregate weighted average net default to claim rate assumption for our primary loans, which is net of estimated denials, rescissions and reinstatements, used in estimating our reserve for losses was approximately 50% at September 30, 2013, compared to 49% at December 31, 2012. As of September 30, 2013, our aggregate weighted average default to claim rate estimate on our primary loans, net of denials, rescissions and reinstatements and excluding pending claims, was 41%, and ranged from 22% for insured loans that had missed two to three monthly payments to 49% for such loans that had missed 12 or more monthly payments.

Our reserve for losses also includes the impact of our estimate of future rescissions and denials, which remain elevated compared to levels experienced before 2009. The elevated levels of our rate of insurance rescissions and claim denials have reduced our paid losses and have resulted in a significant reduction in our loss reserves. Our estimate of net future rescissions and denials reduced our loss reserves as of September 30, 2013 and December 31, 2012 by approximately $291 million and $455 million, respectively. Conversely, our estimate of future reinstatements of previously rescinded policies and denied claims increased our loss reserves as of September 30, 2013 and December 31, 2012 by approximately $286 million and $303 million, respectively, primarily reflected in our IBNR reserve estimate as further described below. The amount of estimated rescissions and denials incorporated into our reserve analysis at any point in time is affected by a number of factors, including not only our estimated rate of rescissions and denials on future claims, but also the volume and attributes of our defaulted insured loans, our estimated default to claim rate and our estimated claim severity, among other assumptions. Although we expect the amount of estimated rescissions and denials embedded within our reserve for losses to remain elevated as compared to levels before 2009, we expect them to continue to decrease over time, as the defaults related to our legacy portfolio decline as a proportion of our total default portfolio and as we realize the results through actual rescissions and denials, or the commutations of insured loans. In the event that we experience a more rapid than expected decrease in the level of future insurance rescissions and claim denials from the current levels, it could have a material adverse effect on our paid losses and loss reserves.

Our reported rescission and denial activity in any given period is subject to challenge by our lender and servicer customers. We expect that a portion of previously denied claims will be resubmitted with the required documentation and ultimately paid; therefore, we have considered this expectation in developing our IBNR reserve estimate. Our IBNR reserve estimate was $268 million and $323 million at September 30, 2013 and December 31, 2012, respectively. As of September 30, 2013, our IBNR reserve estimate of $268 million includes an estimate of future reinstatements of previously denied claims and rescinded policies of $173 million and $78 million, respectively. These reserves relate to $381 million of claims that were denied within the preceding 12 months and $450 million of policies that were rescinded within the preceding 24 months, as well as additional denials and rescissions that were denied or rescinded in earlier periods but remain the subject of discussion with certain of our lender and servicer customers.

The following table illustrates the amount of first-lien claims submitted to us for payment that were rescinded or denied, for the periods indicated, net of any reinstatements of previously rescinded policies or denied claims within each period:



	Three Months Ended September 30,				Nine Months Ended September 30,
(In millions)	2013		2012		2013		2012
Rescissions	$	21.6	$	95.8	$	56.5	$	193.7
Denials	4.1		131.9		131.3		537.1
Total first-lien claims submitted for payment that were rescinded or denied (1)	$	25.7	$	227.7	$	187.8	$	730.8

______________________


(1)	Includes an amount related to a small number of submitted claims that were subsequently withdrawn by the insured.

Generally, we estimate our claim liability related to the potential future reinstatement of previously rescinded policies and denied claims by estimating an initial gross reinstatement rate at the time of denial or rescission, which then declines over a 12 or 24 month time frame. As of September 30, 2013, for previously denied claims, this initial gross reinstatement assumption begins at approximately 60% and declines to 0% after 12 months, while for previously rescinded policies, the initial assumed reinstatement rate begins at approximately 20% and declines to 0% after 24 months. Our IBNR reserve estimate also includes the projected potential impact from future estimated rescissions on reinstated denials and the estimated impact of future claim curtailments on reinstated denials and rescissions. Therefore, at any particular point in time, our IBNR reserve estimate with respect to previously rescinded policies or denied claims is affected not only by our initial reinstatement assumption, but also by the length of time since the denial or rescission, our estimated likelihood of such reinstatements resulting in a paid claim, the expected claim curtailments on such paid claims, as well as the potential outcome of any discussions with our lender and servicer customers regarding such rescissions or denials.

We also accrue for the premiums that we expect to refund to our lender customers in connection with our estimated insurance rescission activity. Our accrued liability for such refunds, which is included within accounts payable and accrued expenses on our condensed consolidated balance sheets, was $24 million and $48 million as of September 30, 2013 and December 31, 2012, respectively.

Rescission and denial rates since 2011 have been affected by an increase in the number of claims received that we review for potential violations of our insurance policies. The following table shows the cumulative rescission/denial rates in our total first-lien portfolio, net of both actual and expected reinstatements, as of September 30, 2013, with respect to claims received in each quarter indicated below:



Claim Received Quarter	Projected Net Cumulative Rescission/Denial Rate for Each Quarter (1)	Percentage of Total Claims Resolved (2)
Q1 2011	20.4%	99%
Q2 2011	25.4%	99%
Q3 2011	30.9%	99%
Q4 2011	27.0%	99%
Q1 2012	23.5%	98%
Q2 2012	20.7%	97%
Q3 2012	17.8%	89%
Q4 2012	16.0%	78%
Q1 2013	16.4%	57%

______________________

(1)

Projected net cumulative rescission/denial rates represent the ratio of claims rescinded or denied to claims received (by claim count). Rescissions and denials are net of actual reinstatements, plus our current estimate for expected reinstatements of previously rescinded policies or denied claims (excluding certain potential reinstatements we are in the process of discussing with servicers). These projected amounts represent the cumulative rates for each quarter as of September 30, 2013. Until all of the claims received during the periods shown have been internally resolved, the rescission/denial rates for each quarter will be subject to change; these rates also will remain subject to change based on differences between estimated and actual reinstatements of previously rescinded policies or denied claims.

(2)

The percentage of claims resolved for each quarter presented in the table above represents the number of claims that have been internally resolved as a percentage of the total number of claims received for that specific quarter. A claim is considered internally resolved when it is either paid or it is concluded that the claim should be denied or rescinded, though such denials and rescissions could be challenged and potentially reinstated or overturned. For the second and third quarters of 2013, a significant portion of claims received for those quarters have not been internally resolved; therefore, we do not believe the projected net cumulative rescission/denial rates for those periods are presently meaningful.

We considered the sensitivity of first-lien loss reserve estimates at September 30, 2013 by assessing the potential changes resulting from a parallel shift in severity and default to claim rate. For example, assuming all other factors remain constant, for every one percentage point change in primary claim severity (which we estimate to be approximately 28% of unpaid principal balance at September 30, 2013), we estimated that our loss reserves would change by approximately $63 million at September 30, 2013. For every one percentage point change in pool claim severity (which we estimate to be approximately 45% of unpaid principal balance at September 30, 2013), we estimated that our loss reserves would change by approximately $3 million at September 30, 2013. For every one percentage point change in our net overall default to claim rate (which we estimate to be approximately 47% at September 30, 2013), we estimated an approximate $39 million change in our loss reserves at September 30, 2013.

Estimating the loss reserves in both our mortgage insurance and financial guaranty business segments involves significant reliance upon assumptions and estimates with regard to the likelihood, magnitude and timing of each potential loss, including the impact of future reinstatements of previously rescinded policies and denied claims. The models, assumptions and estimates we use to establish loss reserves may prove to be inaccurate, especially during an extended economic downturn or a period of extreme market volatility and uncertainty, such as we experienced during the past several years. Our reserves could be impacted in the future if the U.S. economy fails to fully recover from the most recent recession and prolonged economic downturn, including if high unemployment rates continue, or if there is uncertainty in the housing, municipal, foreign sovereign and related credit markets, which could increase our mortgage insurance or financial guaranty losses beyond our existing expectations. As such, we cannot be certain that our reserve estimate will be adequate to cover ultimate losses on incurred defaults.

See Note 9 for information regarding our financial guaranty net claim liabilities.