Non Agency RMBS: The TCW Advantage

The Non Agency RMBS asset class is still ripe with alpha generating opportunities and attractive loss adjusted yields. Nonetheless, the return prospects have come down versus what we have witnessed over the last few years. Yields are lower and spreads are tighter. However, the fundamental trends continue to improve, which should lead to stronger cash flows and total returns for the non-agency market overall. Although the rising tide may have been able to lift all boats in the past, there are cracks and leaks that, if not identified, will lead to underperformance for managers who are not at the analytical forefront of proper bond selection. In a series of white papers we will identify some of the aspects of the market that need to be carefully analyzed in order to capture the most upside while avoiding the many risks that still remain.

Part 1: A Quantitative Approach to Servicer Analysis

When it comes to servicing non-agency mortgage loans, a servicer’s decisions can significantly impact bond cash flows. At TCW we’ve crafted a quantitative servicer ranking system to identify relative strengths and weaknesses among servicers across sectors and capital structure positioning. Some servicing strengths actually become weaknesses depending upon a bond’s placement in the deal capital structure. Consequently, within each sector each servicer receives three distinct rankings, one for each of the following three capital structure positions: Pass Thru, Front Pay, and Back Pay. These scores are derived exclusively from loan level data and are updated monthly1.

Our quantitative approach begins by identifying the decisions a servicer can make that have the largest ability to influence the cash flow to the bondholder. Servicers control the advancing decision, the modification decision, the liquidation decision, and their ability to convince distressed borrowers to send in payments despite being delinquent. Ultimately, a servicer’s relative strengths and weaknesses within these metrics will impact the yields of non-agency bonds differently depending upon where the bond sits in the deal’s capital structure. A simple servicing score could be constructed by adding each servicer’s ranking across the metrics that describe the areas in which a servicer can most influence the cash flow. However, some decisions will have a larger impact on yields than others. For example, the decision whether to modify a loan has more of an impact than the number of months it takes to modify (see below). Hence, after generating rankings across our servicing metrics for the servicers, we weight those rankings by the potential yield impact of each metric. The more a decision can impact a bond’s yield the larger the weight it carries in our servicer scoring system.

Table 1 shows the weights to the metrics we use to rank servicers across sectors and capital structures. After ranking relevant servicers within each metric, these numerical rankings are subsequently weighted by the percentages shown in Table 1 to produce a metric score. Scores across all metrics are summed to get a servicer level score. We use this methodology to score servicers by sector and by capital structure. What follows is an analysis of each of these servicing metrics and the potential yield impact used to generate the weights in Table 1.

Table 1: Servicing Metric Weights2

Source: TCW, Corelogic, Intex, RBS


The decision to advance delinquent payments for the borrower clearly is the most influential metric on our list. Typically a servicer for a non-agency securitization is required to advance delinquent payments until it deems them “unrecoverable.” Servicers have ample discretion when deciding what “unrecoverable” means. During a distressed liquidation, numerous costs eat away at the potential recovery of funds from the sale of the home. These costs include attorney fees related to foreclosure action, eviction costs, sales commissions, property maintenance, property taxes, insurance, and servicer advances of delinquent payments, etc. If the servicer believes that the proceeds from the sale of the home, net of costs, are at risk of being insufficient to cover its advances, it will stop advancing.

The decision to stop advancing is typically characterized as a cash flow neutral event since, at the pool level, investors either receive the cash today as an advance (and have it taken out of the liquidation proceeds by the servicer later) or forego the cash today in exchange for a lower loss severity in the future (since the servicer will no longer have claim on any previously advanced funds).

For example, if a bondholder holds the current cash flow bond in a sequential pay capital structure and the servicer stops advancing payments for delinquent borrowers, then the bond will extend its duration and, depending upon the magnitude of the stop-advancing behavior, could take principal losses that would not have been a risk otherwise. However, since the Back Pay bond has a longer duration and only receives bond coupon cash flow as the Front Pay pays down, it benefits from the lower severities expected a few years out after the Front Pay is gone. Hence, the advancing decision has less of an influence in the loss adjusted yield of a Back Pay bond.

Figure 1 shows the percentage of seriously delinquent Subprime unpaid loan balance that servicers have stopped advancing upon.

Figure 1: Subprime Percentage 60+ Not Advanced Upon by Servicer

Source: TCW, Corelogic, RBS

When isolating the impact of the stop advance decision on our three types of bonds, we use the range of stop advancing found among the Subprime sector. On one end of the spectrum, there are pools where advances occur on 97% of the seriously delinquent collateral. On the other end of the spectrum we find pools that have only 15% of the seriously delinquent loans with advances. As we model these cash flows, we keep the severity the same over time for loans where the servicer advances delinquent payments, and we have the severity decrease over time for loans where the servicer has stopped advances. We model these differences using our three types of bonds: Pass Thru, Front Pay, and Back Pay. The results indicate that this 82 percentage point difference in advancing behavior and relevant severity ramps results in yield differences of 268 bps, 492 bps, and 175 bps, respectively. By far, for a Front Pay bond, the most influential decision a servicer can make is whether or not to advance. Under this approach, our analysis shows that the more advancing that occurs, the higher the yields across the capital structure all else equal.

Modification and Recidivism

Loan modifications come in all shapes and sizes. The modification that we care about most when evaluating servicers is the modification that cures delinquency permanently. If a borrower on a modified loan subsequently falls back into delinquency (recidivism) and the property is eventually sold at a loss then the modification only served to delay the inevitable liquidation. When it comes to modifications, a good servicer will be able to identify the borrowers most likely to remain current post-modification and not waste time (and carry) on borrowers who will become recidivist delinquencies. This usually involves matching the right kind of modification with each borrower. If a servicer were able to successfully modify all the delinquencies in the pool then cumulative default expectations would drop dramatically and in most cases total returns would increase for bondholders. Unfortunately, there are no servicers who have successfully modified all their delinquent borrowers. But some servicers have modified more loans than others and some servicers are more successful at modification than others.

We stress our three example bonds across the existing ranges of modification and recidivism rates available in the industry. To do this, we assume that borrowers who have never missed a payment are excluded from the pool of potential modification candidates. When we examine non-agency 2006-2007 Subprime loan pools we find a broad range of modification activity. At one end of the spectrum we find pools where only 20% of the loans that have been in delinquency at least once have been modified. On the other end of the spectrum we find pools where 94% of the loans that have been in delinquency at least once have been modified. Next, we take the pipeline of serious delinquent loans for each of our three example bonds and modify 20% of those loans never modified and rely upon an industry average recidivism rate to project final delinquency post modification activity. This leads us to a cumulative default projection for the pool. We repeat this process in another stress modifying 94% of all previously unmodified delinquent borrowers in the pool. The yield impact of these stresses on the Pass Thru, Front Pay and Back Pay are 196 bps, -16 bps, and 85 bps, respectively. Front Pays have a strong bias for servicing behavior that generates current cash flow. A modification stops a large liquidation producing cash flow event in favor of significantly smaller monthly payments from the modified borrower. Consequently, a modification has a slightly negative impact on the yield expectations for a Front Pay despite bringing down the cumulative default expectation.

To stress recidivism we assume that all 94% of our non-modified delinquent loans are modified and allowed to return to delinquency at rates defined by the two ends of the recidivism spectrum: 23% and 66%. That is, the most successful example of modification has a 23% recidivism rate and the worst servicing performance in this category has a 66% recidivism rate. These modifications and recidivism rate stresses result in reduced cumulative default expectations, which when translated into yield impact generate 122 bps, -12 bps and 112 bps of additional yield for the Pass Thru, Front Pay, and Back Pay, respectively. Again, because of the Front Pay’s bias towards current cash flow, the recidivism component of servicing behavior has a slightly negative impact on the lossadjusted yield.

Cash Flow Velocity

Recognizing that even the best servicers require several months to modify and liquidate delinquent loans, we incorporate cash flow velocity into our evaluation. Even though a borrower may be 60 days delinquent, a skilled servicer has the ability to contact the borrower and convince him to make a single payment so that the loan does not advance to 90 days delinquent. Cash flow velocity is the percentage of the monthly balance due that a servicer was able to obtain on loans that are already delinquent. This statistic ranges between 30% and 49% in the Subprime universe (Figure 2).

Figure 2: Cash Flow Velocity by Servicer

Source: TCW, Corelogic, RBS

That is, the worst servicer is able to extract 30% of the monthly balance due from delinquent borrowers and the best servicer in this category is able to extract 49% of the monthly balance due. This range translates into 68 bps, 187 bps, and 54 bps, of yield for Pass Thru, Front Pay, and Back Pay bondholders, respectively.

Loss Mitigation Timelines

A servicer’s decisiveness also plays a role in our ranking system. These days, a servicer has two choices when a borrower becomes delinquent. It can modify the loan or it can liquidate the loan. Whatever choice is made, the best servicers will implement the solution as quickly as possible and minimize modification timelines and liquidation timelines. The more payments that a borrower misses the higher the loss is to the investor. Servicers with the fastest timelines to modify and to liquidate score higher in our ranking system. Figure 3 shows the range of months to liquidate by servicer.

Figure 3: Subprime Avg # Months from DQ to Liquidation

Source: TCW, Corelogic, RBS

Figure 4 shows the range of months required to modify by servicer.

Figure 4: Subprime Avg # Months in DQ Prior to Modification

Source: TCW, Corelogic, RBS

When we stress our bond cash flows by the impact that these ranges of timelines can have on yields we find that liquidation timelines result in 116 bps, 47 bps, and 46 bps of additional yield for Pass Thru, Front Pay, and Back Pay, respectively. Additionally, the impact that the ranges of modification timelines can have on yields is 21 bps, 8 bps and 9 bps of additional yield for Pass Thrus, Front Pay, and Back Pay, respectively.

As mentioned above, we compile a servicer’s score by assigning a rank to each servicer for each metric (Advancing, Modification Volume, Recidivism, Liquidation Timeline, Modification Timeline, and Cash Flow Velocity). Then we weight each ranking by its ability to impact the yield (Table 1). The weighted sum of each ranking produces an overall servicer score for each capital structure type within each sector. This score is updated each month as new loan data become available. For Pass Thru’s and Back Pays we like servicers who advance on delinquent loans, emphasize successful modifications, have high cash flow velocity and are quick to modify or liquidate. For Front Pays, current cash flow is essential. Above all, a good Front Pay servicer will emphasize advances and cash flow velocity. Modification and recidivism are slightly detrimental to yield and fast timelines add a few basis points of yield. See our current servicer rankings by capital structure and sector below in Tables 2, 3, and 4.

As we navigate the landscape of lower yields in the non-agency sector, understanding how servicing behavior can impact a bond's yield becomes increasingly relevant. Our approach to matching servicing behavior to sectors and cash flow types gives TCW a distinct advantage to methodically construct alpha generating portfolios.

Table 2: Servicer Rankings for Pass Thrus3

Source: TCW, Corelogic, RBS

Table 3: Servicer Rankings for Front Pays

Source: TCW, Corelogic, RBS

Table 4: Servicer Rankings for Back Pays

Source: TCW, Corelogic, RBS

1 Monthly vigilance is required in the servicing industry due to a rising trend in servicing transfers and consolidation.

2 Green represents a positive impact on yield. Red represents a negative impact on yield.

3 Our servicing list covers the largest servicers in the industry with a minimum of at least 1 billion in loans serviced.

Any issuers or securities noted in this document are provided as illustrations or examples only, for the limited purpose of analyzing general market or economic conditions, and may not form the basis for an investment decision. TCW makes no representation as to whether any security (or the security of any issuer) mentioned in this document is now or ever was held in any TCW portfolio. TCW is not recommending the purchase, sale or holding of any security and is making no representation or indication of its own holdings of any securities. TCW may in fact be currently recommending the purchase of a security or the sale of a security regardless of any statement made in this document about that security or whether TCW owns it or not. Discussion of securities in this document are strictly for educational use only and are not intended to serve as investment advice. Any statement made in this document, including any statement or implication drawn from any discussion of individual securities, is subject to change at any time, without notice.

For Information Only

This publication is for general information purposes only. Past performance is no guarantee of future results. While the information and statistical data contained herein are based on sources believed to be reliable, we do not represent that it is accurate and should not be relied on as such or be the basis for an investment decision.

Subject to Change

Any opinions expressed are current only as of the time made and are subject to change without notice. TCW assumes no duty to update any such statements. The views expressed herein are solely those of the author and do not represent the views of TCW as a firm or of any other portfolio manager or employee of TCW. Any holdings of a particular company or security discussed herein are under periodic review by the author and are subject to change at any time, without notice. In addition, TCW manages a number of separate strategies and portfolio managers in those strategies may have differing views or analysis with respect to a particular company, security or the economy than the views expressed herein.

MetWest is a wholly-owned subsidiary of The TCW Group, Inc.

© TCW Asset Management

Read more commentaries by TCW Asset Management