In recent years, the non-Agency residential mortgage-backed securities (MBS) sector has been a strong source of returns for many investors, and we believe the sector continues to offer value. However, extracting alpha away from beta from non-Agency MBS requires careful evaluation of the risks associated with this complex asset class, including the future trajectory of home prices, public policy, capital structure, liquidity and mortgage servicing.
What makes investing in this asset class so complex? First, unlike Agency MBS, non-Agency securities are backed by non-government guaranteed mortgage loans. While the primary concern for an Agency MBS investor is when principal will be returned (prepayment risk), non-Agency MBS investors have two primary concerns: both when and if principal will be returned (prepayment and credit risk). In addition, non-Agency MBS deals consist of several different securities, or “tranches,” each with its own degree of exposure to credit, prepayment and liquidity risks. Given the presence of credit risk in these securities, cash flow modeling is key to understanding how cash flows and losses are distributed throughout the deal structure.
Granular housing analysis is critical
Typically, mortgage borrowers are likely to prepay their loans if they have equity in their homes and an economic incentive to refinance. They are also less likely to default if they have equity in their property. To forecast borrowers’ prepayment and default propensity, you need to have a view on the trajectory of home prices. And while many people view the U.S. housing market as one large market, it is in fact highly localized (see Figure 1). Beyond differences from state to state and city to city, there are even large discrepancies in home price performance from zip code to zip code (Figure 2).
This means that an investor’s view on the trajectory of home prices needs to be extremely granular. To develop a forecast on national home prices, PIMCO uses macroeconomic conclusions from its quarterly forums, as well as forecasts tailored by zip code, based on key factors including local income and unemployment trends, market dynamics (foreclosure laws, backlog of distressed supply), data and anecdotal information from our non-performing loans business and local market qualitative research (PIMCO “ride-alongs”). Much of the well-publicized housing data is reported on a several month lag, and only focuses on the larger U.S. cities. Many housing indices are also based on recent transactions, which can result in material volatility in data points as index composition changes from distressed to non-distressed transactions. For all of these reasons, PIMCO’s proprietary local home price forecasts are a key factor in our non-Agency MBS security selection process.
Analyzing public policy factors
Public policy has played a key role in the non-Agency MBS market in the post-crisis environment, both in the form of accommodative monetary policies and innovative housing policies. Fed-induced low mortgage rates have driven refinance and purchase activity in the Agency market, supporting housing overall. In addition, loan modification programs such as the government-advocated Home Affordable Modification Program (HAMP) affect the amount of cash flows that will be available to bondholders.
Therefore, policy analysis, including forecasts on Agency conforming loan standards, the pace of Fannie Mae and Freddie Mac risk reduction and the path of housing finance reform, is also an integral part of PIMCO’s investment process. At PIMCO, we rely on several public and housing policy experts to determine the potential impact of policy-related factors on non-Agency MBS valuations.
Non-Agency MBS deals typically comprise a variety of “tranches” with highly differentiated risk/return characteristics based on their position in the capital structure. Typically, investors use third-party cash flow models to value various bonds within the same deal and to determine trade-offs across the credit and maturity spectrum. However, the cash flows of non-Agency MBS deals can be highly complex, and investors have no guarantee that the third-party cash flow distribution model is completely accurate. At PIMCO, we carefully analyze deal documents to determine exactly how cash flows will be distributed to the various bondholders in the deal structure. At times, discrepancies in deal documents and/or cash flow distribution models can be a source of both risk and opportunity. Therefore, careful analysis of the cash flow position of each tranche in these deal structures (or “sum of the parts” analysis) can often determine which securities may have the most attractive risk/reward profile in a given non-Agency MBS deal.
Liquidity can be limited, as much of the sector lacks investment-grade credit ratings.
Based on our data, $830 billion of non-Agency MBS were outstanding as of 30 November 2013, down from approximately $2.1 trillion at the peak of the housing market. Of that, approximately 93% percent of bonds originally rated AAA are now rated below investment grade (see Figure 3). This limits the demand base for this asset class, as many fixed income investors are restricted in their ability to purchase below-investment-grade securities. While many non-Agency MBS are rated CCC by rating agencies, they likely would carry higher ratings if they were rated through a methodology similar to that for corporate bonds. For example, if a non-Agency bond is expected to take one point of principal loss, then it is likely to be rated CCC by the rating agencies, whereas a corporate bond with a similar expectation of loss will likely still be rated investment grade. PIMCO assigns internal credit ratings to bonds on an expected-loss basis and compares the risks and rewards with those of other credit sector alternatives.
Positioning on the credit curve
As an investor, one needs to decide what the incremental spread compensation is in a high quality versus low quality non-Agency MBS bond. In addition, investors should determine what the incremental compensation is within a “quality bucket” for two-year spread duration risk versus 10-year spread duration risk. A credit curve helps determine trade-offs between credit quality and spread duration.
Depending on portfolio objective and macroeconomic outlook, an investor may favor different parts of the non-Agency credit curve. At times, we may favor the front end of the credit curve, where there is relatively little credit risk and relatively high compensation in terms of loss-adjusted spread. Other times, we may prefer lower-quality bonds in the intermediate part of the curve (six- to nine-year spread duration), where we give up loss-adjusted spread compensation per unit of spread duration risk but gain upside price appreciation potential as the market shifts from weaker to better housing scenarios. Based on our specific portfolio strategy, we constantly make relative-value decisions on our credit curve positioning by evaluating the trade-off between financial and structural leverage.
Mortgage servicing matters
Mortgage servicers are key players in the MBS sphere. They are responsible for collecting payments from borrowers, liquidating delinquent loans and administering loan modifications to borrowers to maximize value to the trust. Servicers also typically advance (make payments) on delinquent loans when they believe these payments will be recovered. Advances can be a key source of cash flow for non-Agency MBS investors, especially in scenarios with negligible prepayments and defaults.
However, no two servicers are alike when it comes to deciding whether to modify a loan. Figure 4 shows the 60-plus-day delinquency trend for 2006 vintage subprime loans across several servicers. Over time, the difference in servicing practices is a likely factor behind diverging performance. In addition, conflicts of interest can arise between investors and servicers – and resolution of these conflicts tends to be poorly defined, if at all. In an extreme hypothetical scenario, a servicer may try to extend the stream of fees it earns by modifying a loan instead of liquidating it. A servicer could also be inadequately staffed, trained or incentivized to perform “high-touch loan modifications” (modifications that require a high degree of interaction with the borrower) on high-delinquency pools, leading to loss of value for investors. Finally, recent mortgage servicing right (MSR) sales from large banks can make it harder to forecast cash flows due to uncertainty regarding changes in process for advances, modifications and liquidations. Therefore, PIMCO’s investment process considers a wide range of servicing issues when forecasting cash flows.
Consider every risk factor when investing in non-Agency MBS
Non-Agency MBS can potentially be an attractive source of risk-adjusted returns, but a multifaceted investment approach focused on detailed security selection and strong downside mitigation is critical. At PIMCO, we employ over 50 mortgage investment professionals, focused on everything from granular housing market analysis to servicer surveillance, in order to ensure that we identify any hidden risks or opportunities within the sector. We believe the
non-Agency MBS market continues to present attractive value relative to most traditional fixed income sectors, especially for investors who can properly analyze the myriad risk factors within these securities.