While a number of factors beyond a servicer’s control -- including dips in the economy and various characteristics of the individual borrowers -- make it difficult to predict defaults and recoveries directly, Standard & Poor’s Rating Services’ Global Surveillance Analytics team has developed a method for assessing a servicer’s default management process. Investors can use these methods to assess how effective servicers’ strategies are in limiting losses.
Our new method incorporates two key components: how quickly individual servicers have been able to liquidate nonperforming loans; and the success of their loan modification programs. We estimate the net effect of these aspects of servicers’ strategies on losses into a single number that reflects the cost or benefit of the timing differences and the loss reductions and costs related to modifications. The results of our study suggest that servicers’ strategies can greatly affect losses resulting from nonperforming loans. They also reveal significant performance variations among the servicers in our sample.
This method is separate from that of Standard & Poor’s Servicer Evaluations group, which encompasses a much broader set of factors for assessing and ranking servicers. For this study, we used loan-level information on nonagency U.S. residential mortgage-backed securities from CoreLogic’s LoanPerformance database, assigning each loan to its respective servicer where possible (many loans remain unassigned). The loans included are not limited to deals we rate, but include all securitized first-lien nonagency loans available through LoanPerformance. Although we believe we identified a substantial number of loans for many of the largest servicers, the data set is not complete. As such, we do not list the servicers by name (although we may do so in future reports) and limit the results to the ten servicers for which we could identify the largest number of loans.
Metrics For Assessing Servicers’ Resolution Timelines
We view the average amount of time to close a nonperforming loan -- which includes liquidations, as well as deeds-in-lieu of foreclosure and short sales -- as a particularly useful method for assessing servicers, for several reasons. Long foreclosure periods typically increase losses due to missed interest payments and incurred expenses. Moreover, because we can easily conduct our analysis to account for the most significant confounding factors -- the loan’s pool type and the location of the property -- the time to close a nonperforming loan provides a more accurate and less subjective measure, in our view.
To assess servicers’ efficiency, we looked at the average number of missed mortgage payments for each servicer’s nonperforming loans (including loans that are still open and those that closed after they defaulted), then subtracted any additional mortgage payments the borrower made while delinquent. This provided a rough measure of the relative cost to an investor of the servicer’s resolution, as accrued interest is the largest controllable component of loss severity.
Average foreclosure timelines -- and, consequently, the number of missed mortgage payments before liquidation -- can vary significantly between states, primarily because of variations in their particular judicial foreclosure requirements. Some states, such as New York, require a lengthy mediation process that can add several months to liquidation timelines (previously discussed in our report, “New York Can’t Get No Liquidation,” 4/12/11).
For each servicer, we calculated the difference between the average actual number of missed payments (AMP) on its loans and a target number of missed payments (TMP). A servicer’s TMP is an overall average number of missed payments that accounts for the distribution of the servicer’s loans across states and loan types and variations in the average number of missed payments by state and pool type.
To calculate each servicer’s TMP, we first divided the mortgages from all servicers into 153 groups: 51 regions (50 states plus Washington, D.C.), with three loan types for each (prime jumbo, subprime, and Alternative-A). We then calculated the average number of missed payments for each of these 153 groups. Each servicer’s TMP is a weighted average of the average number of missed payments in each of the groups, weighted by the number of loans that servicer monitors in each group. This target, shown on the left-hand-side of chart 1, represents an approximation of what each servicer’s average number of missed payments would be if the servicer closed its nonperforming loans as quickly as the average in each of the 153 groups of loans. On the right side of chart 1, the AMPs show the actual average number of missed payments for each servicer’s closed loans. The difference between a servicer’s AMP and TMP, or the steepness of the line in chart 1, is a measure of how quickly a servicer closed its nonperforming loans relative to the average after accounting for the distribution of the servicer’s loans across regions and pool types. A servicer’s line going up in chart 1 (from left to right) means that the servicer resolves its nonperforming loans more slowly than expected, and vice versa.
We calculated an AMP and TMP for each servicer for two groups of loans: loans that were more than 90 days delinquent but still active as of November 2011 and defaulted loans that the servicer had closed between January 2009 and November 2011. As in chart 1, the higher the number, the longer the liquidation timeline relative to the peer average, and vice versa.
We found significant differences in servicers’ speeds relative to their targets. For instance, servicer B was, on average, able to close its nonperforming loans 3.4 months more quickly than its target, saving investors that many months of missed interest payments and preservation costs. On the other end of the spectrum, it took servicer J 3.9 months longer than its target to close its nonperforming loans. This means that just among the 10 servicers in our sample, investors might be able to save up to 7.3 months of interest payments (on loans that eventually default and close) by discriminating between servicers.
Metrics For Assessing Servicers’ Loan Modifications
Servicers may also have the option to modify the terms of certain loans to make it easier for the borrower to pay and thus less likely to redefault. We consider a modification successful if it enables a delinquent borrower to start making payments again and the loan remains current for at least 12 months.
Since successful modifications can result in investors avoiding significant losses, both the magnitude and relative success rates of servicers’ loan modification programs are relevant in analyzing servicers’ strategies. We found that for the servicers in our sample, rates of successful modifications as a percentage of all nonperforming loans ranged from 3% to 18%.
Making Sense Of The Numbers
As a final step in our analysis, we combined the servicers’ relative resolution timelines with their modification success and failure rates. To do this, we came up with a method that converts each factor into an estimated effect on investor losses. For this aggregate calculation, we assume that each month difference between a servicer’s AMP and TMP saves interest payments and preservation costs of about 1%.
For loans successfully modified, we assume servicers have avoided 75% of otherwise expected losses on those loans, reflecting the costs associated with modifying a loans and the fact that the loan may eventually redefault. Since loss severity averaged 60% for the pool of loans in our sample, a successful modification implies savings of 45% (75% times 60%) for investors. On the other hand, failed modifications can be costly as they generally restart the foreclosure process--when the loan redefaults, it will incur additional lost interest and other expenses. We assume that this delay in ultimately resolving the loan, along with the direct costs of implementing the modification, result in additional losses of 9%.
The results of this calculation are shown in the last column of table 1. Using these assumptions, servicers can have a considerable effect on reducing the losses investors suffer from nonperforming loans. In our study sample, the effect ranged from additional losses of 3% (represented by servicer J’s -3%) to an 8% reduction in losses (represented by servicer A’s +8%). This range implies that servicers’ decisions can have significant impact on the amount of losses investors experience.
We believe the methods developed in this study provide a solid base for analyzing some key aspects of servicers’ success in reducing total losses to U.S. RMBS investors. The combined number we calculate allows investors to compare the various elements of servicers’ strategies in dealing with nonperforming loans to estimate the net effect on losses. In the coming year, we plan to use loan-specific information to refine our estimates of the effects of servicers’ resolution timelines and modification programs. In addition, we intend to create a model to grade servicers’ ability to prevent defaults from occurring in the first place.
We believe these metrics will be useful for investors looking to assess the relative performance of servicers’ strategies. Our approach captures significant differences between servicers that can have considerable impact on losses. Servicers that can close nonperforming loans more quickly save investors several months of missed interest payments, while extensive loan modifications, when successful, can prevent losses.