Mortgage-backed securities (MBS) are often overlooked as a source of long-duration exposure in corporate defined benefit plan portfolios, where long government and long credit bonds typically dominate. Yet agency MBS offer one of the highest long-term information ratios relative to U.S. Treasuries and the lowest correlation of excess returns to the S&P 500 Index, as we detailed in “The Unique Benefits of Mortgage-Backed Securities.” In this Q&A, we discuss how a portion of the MBS market is an attractive fit for long-duration investors.

Q: Long corporate bonds would seem to be natural hedging assets in a liability-driven investing (LDI) portfolio because they are used to calculate the discount rate for pension liabilities. How do MBS work in this context?

Q: Long-dated liabilities have high positive convexity while agency MBS are negatively convex instruments. How can they play a role in long-duration fixed income portfolios designed to hedge pension plan liabilities?

However, the process of structuring agency MBS cash flows creates a series of collateralized mortgage obligations (CMOs). Instead of tranching credit risk, where bonds are created with varying degrees of credit risk, agency CMOs tranche prepayment risks, creating classes with varying degrees of exposure to changes in interest rates, the shape of the yield curve and prepayment speeds on underlying mortgage loans. This can dramatically change the cash flow stability and convexity of MBS. Active management can play a further role in identifying security structures and collateral characteristics best suited for liability-matching.

Figure 1 shows how cash flow structuring of a 4% MBS guaranteed by Fannie Mae with a yield of approximately 2% and a weighted average life of 3.7 years can create two securities – a shorter, lower-yielding security and a longer-duration, higher-yielding security. Essentially, this creates a 12-year bond with a yield of 3.2% by limiting the potential extension risk of the three-year bond which yields 1.4%. The shorter bond at the front end of the structure absorbs the majority of initial prepayments, providing a locked-out period where the longer bond is insulated from prepayment risk. While duration- sensitive investors may seek to purchase the shorter bond, the longer-duration bond with superior convexity creates an additional tool in the long-duration toolkit for liability-hedging plan sponsors.

Q: How should investors think about an allocation to long-duration agency MBS?

We believe that long-duration agency MBS work best as part of a broader long-duration portfolio, which presumably focuses on long-duration government bonds and/or long corporate credit.

Historically, the role of long government bonds in LDI has been to diversify the long credit component of the fixed income portfolio and/or equity exposure at an overall asset allocation level. In essence, this is a barbell approach to diversification, with government bonds and corporate bonds at opposite ends of the credit risk spectrum. Yet, although diversification benefits are susceptible to changes in correlations, a barbell approach can magnify risks.

A more balanced approach would prudently allocate long-duration CMO exposure across a broader long-duration portfolio. For example, long-duration MBS can be used to gain additional spread relative to long-duration U.S. Treasuries, while maintaining a high-quality bias with little corporate credit risk. Instead of earning compensation for incurring credit risk in long-duration credit, investors are paid for the cash flow uncertainty associated with prepayment risk in agency MBS.

As Figure 2 shows, yields of long-duration MBS are greater than the Long Government Index, with less duration. Long-duration MBS yields are significantly lower relative to long credit, especially given the recent cheapening in corporate credit risk. Importantly, as of 29 February, the correlation of long MBS to long government bonds was 0.75, while the correlation to long credit was just 0.59. However, while long-duration MBS are U.S. government/agency-guaranteed, they carry spread duration; in risk-off, flight-to-quality rallies in interest rates, MBS spreads can widen materially, resulting in mark-to-market volatility on long-duration MBS relative to long-duration U.S. Treasuries.

We take this dynamic into account in scaling long-duration MBS exposures in portfolios. In full-discretion portfolios, we typically allocate up to 10% after careful analysis of the trade-offs between enhanced return potential and tracking error relative to liabilities. Understanding how and when to scale exposure to this asset class can potentially mitigate the overall convexity risk of a portfolio.

Q: High-quality corporate bonds are often limited in supply. What’s the size of the market for long-duration CMOs?

Q: What are the risks in a long-duration MBS portfolio?

Liquidity risk: Government-guaranteed, long-duration MBS do not trade with nearly the same bid/ask spreads or daily trading volume as long-duration U.S. Treasury or agency bonds. This can create unexpected volatility, as was witnessed in 2008 when agency MBS spreads widened to 300 basis points (bps) above U.S. Treasuries. While the sector held up significantly better than investment grade corporates (where spreads exceeded 500 bps), it is an important example of agency MBS not always serving as a “hard” duration asset. The sector can be prone to material dislocations in liquidity, especially as it relates to CMOs.

Convexity/prepayment risk: Long CMOs can provide significantly more cash flow stability/prepayment protection than a traditional agency MBS pass-through, but only to a degree. When prepayment speeds increase significantly and for an extended period, bonds with payment priority ahead of the long CMOs may amortize faster than expected, resulting in a declining cash flow buffer to protect the long CMO from increasing prepayment speeds. Over time, this can result in a major shortening of the long CMO cash flow, creating mismatches for liability-driven investors.

Given that the sector’s liquidity can be fragile and scaling can be challenging, we believe long CMOs are most useful as an off-benchmark allocation in a broader long-duration government/credit portfolio. Furthermore, if investors are seeking stand-alone, long-duration MBS strategies, a strong understanding of both prepayment risk and cash flow structuring is critical to long-term success.

Q: Would adding MBS to an LDI allocation increase risk relative to liabilities?

First, liability discount curves were not intended for investing but as a proxy for valuing a pension plan’s liability for accounting or funding purposes. Much has been written about this topic, but the long AA corporate bond universe is generally uninvestable in isolation given severe concentration risk – the top five issuers represent 60% of the index, and supply of bonds in this segment is waning. Additionally, there are many flavors of the AA discount curve with various constraints, provider exclusion rules and spline/curve fitting methodologies, rendering it an abstract representation of liability-matching investments. This argues for improved diversification in an LDI portfolio with a broadened credit universe that includes the full investment grade set of bonds and potentially other fixed income securities such as government bonds and long-duration MBS.

Second, while an LDI portfolio dominated by credit investments can face potential defaults (albeit a rarity in the investment grade space) and downgrades, liabilities cannot default or be devalued due to downgrades; i.e., the plan sponsor is obligated to provide the promised benefits to their plan participants. Thus, we suggest reducing single-name concentration as much as possible, with the constraint that the diversifying asset is not meaningfully under-yielding a duration-equivalent long corporate bond. Given the potential diversification benefit and yield trade-off shown in Figure 2, long-dated mortgages would be preferred over Treasuries to some extent.

Third, stepping back and looking at the broader pension asset allocation, it is important to consider overall asset-liability risk, or more specifically, the drivers of funding-ratio volatility. In a majority of plans, equity allocations still dominate and contribute meaningfully to this risk. The composition of the liability-hedging allocation also affects overall risk, and having a marginal allocation to long-dated MBS may improve diversification at the overall plan level, which in turn would help reduce funding-ratio volatility.

Q: How should a plan sponsor think about implementing a long-duration MBS strategy?

• Are long-duration MBS cheap relative to other long-duration assets?
  
  
• Are there diversification benefits to owning long-duration MBS?

Crucially, if investors are seeking stand-alone long-duration MBS strategies, a strong understanding of both prepayment risk and cash flow structuring is critical to long-term success.

Q: How would PIMCO design a stand-alone, long-dated MBS portfolio?

In addition, select use of agency pass-throughs, longer-dated agency and potentially non-agency collateralized CMBS and even relative-value overlay strategies can expand the opportunity set in a long-duration MBS portfolio. Ultimately, portfolio construction would depend on the level of discretion and ultimate goal of the portfolio (total return versus liability-matching substitute).