The Perils of Underweighting Mortgages

Mortgage-backed securities (MBS) are one of the largest sectors within the bond market, with $6 trillion of market value. This is second only to US Treasuries in broad-based indexes,¹ both in terms of notional value and weight (29%). Despite this size, investment portfolios are decidedly underweight the bond sector—even though its investment profile has historically been attractive. As a result, this mortgage investment reluctance can be a detriment to portfolios, both from a historical perspective and based on the current market dynamics of low rates, extended duration risk, and tight credit spreads.

Mortgage market underweights are broad
In our annual review of investor portfolios, we identified an underweight to the agency and pass-through mortgage sector. While these reviews cover a diverse user base, it is still a small selection of investment capital. To form a broader view on the relative positioning by investment portfolios in MBS, we analyzed holdings of ETFs and mutual funds.

Here are the findings:

• ETFs categorized as mortgage-backed make up only 5% of the overall bond ETF market, much lower than the market share of targeted government and corporate bond ETFs, at 17% and 19% respectively.² In fact, the market share of MBS is the same as inflation-protected strategies, a more niche area of the market.
• One theory could be that investors obtain MBS exposure through core Aggregate-based (Agg) funds. However, combining all 108 strategies classified as “Aggregate,” 52 of which are active mandates, into one omnibus portfolio reveals that MBS comprises only 21%³ of the $280 billion assets.
• As expected, the indexed-based ETFs tracking the Agg have the 29% weight to MBS. However, the 52 active funds in this category, covering $60 billion in assets, have only 13%⁴.
• MBS exposure through the usage of active core bond mutual funds is also underweight relative to the Agg, on average. Out of the $1.8 trillion in assets in the two largest core-oriented Morningstar categories (intermediate core⁵ and intermediate core-plus⁶ ), MBS exposure is 27% and 22%⁷ , respectively.

As this shows, no matter the pool of strategy assets, investor portfolios are likely to underweight MBS relative to the typical reference benchmark: the Agg.

Mortgage behavioral bias
Some of this investment reluctance could be attributed to behavioral biases stemming from the fallout from the Great Financial Crisis (GFC), a time period that sometimes is also referred to as the mortgage crisis. But it wasn’t agency mortgages that were in crisis; it was the non-agency private-label collateralized mortgage obligations that do not have the explicit backing of the federal government (Ginnie Maes) or the implicit backing from the government-sponsored entities (GSEs) Fannie Mae and Freddie Mac. In fact, most investors probably do not realize that agency MBS from 2007–2009, a time period covering both the GFC drawdown and the start of the recovery, returned 20.27%—a figure that is much higher than the broader Agg, investment-grade corporate bonds (IG corporates), US Treasuries, and US equities, all along with the least drawdown, as shown below.

Understanding the differences between the two major types of MBS is helpful in overcoming this behavioral heuristic of mortgage aversion.

Below are the major differences:

• Agency mortgage-backed security: An asset-backed security representing the amount of interest in a pool of mortgage loans, backed by an implicit or explicit guarantee on the timely payment of cash flows—decreasing default risk to a negligible level.
• Non-agency private-label collateralized mortgage obligations: A security divided into categories based on risk and maturity dates that pools mortgages into a special-purpose entity, from which different tranches of the securities are then sold to investors. These are subject to default risk, evidenced by the value of these securities falling from $1.3 trillion in value in 2007 to just $850 million in 2013⁸.

Investment implications of missing the mortgage market
Without any behavioral biases from investors, the aversion to mortgages could be understood if the risk and return profile had been lackluster. However, based on the historical return, correlation, and volatility profiles, that is not the case.

There are five reasons why mortgages may be a valuable component to a portfolio:

1. Yield enhancement: Mortgages have historically earned a premium over Treasuries and the broader Agg. Since 1997, the average yield for MBS was 4.5% compared to 3.2% and 4.1%, respectively⁹ . This average 43% and 11% premium is not only witnessed on a straight-line historical average that can be skewed by outliers, but it is also evident with historical persistency. In 100% of the 264 rolling 12-month periods since 1997, MBS had a higher average yield than both Treasuries and the Agg, respectively, as shown here. Naturally, this profile continues today (2.5% versus 1.8% and 2.3%).

2. Higher yield per unit of duration: The higher yield is not, however, a result of extending on duration. With an average of 3.5 years of duration, MBS have historically had 41% and 25% less duration than Treasuries and the Agg, reductions that are on par with levels today. As a result, MBS have a far superior average yield per unit of duration ratio of 1.29, versus 0.57 and 0.84 for Treasuries and the Agg, respectively. MBS have also had 50% less duration risk than IG corporates, historically. While there is a duration reduction benefit versus IG corporates, MBS do not have the same yield advantage.

3. Lower volatility: IG corporates historically have yielded 8% more than MBS. But that makes sense, as there is more credit risk in IG corporates, and investors should be compensated for bearing that risk. From a yield per unit of risk perspective, however, MBS still wins out. IG corporates’ 8% yield advantage is not enough to make up for the 50% more duration risk—leading to an attractive yield per unit of duration ratio for MBS, as shown above. And the same holds true from a total risk perspective, in addition to just interest rate risk (duration).

Examining the standard deviation of absolute returns (i.e., total risk) reveals that MBS have low relative volatility properties. MBS have 41%, 51% and 25% less volatility, as measured by standard deviation of monthly returns, compared to Treasuries, IG corporates and the broader Agg. As a result, MBS have a yield per unit of standard deviation, or rather risk-adjusted yield, that is 141%, 88% and 47% higher than those three segments, respectively. Downside deviation is also improved, leading to lesser drawdowns historically. The max drawdown for MBS is just -2.8% compared to -4.3%, -15.2%, and -3.8% since 1997.

4. Higher credit quality: The lower volatility is a byproduct of the lower duration (versus Treasuries) and improved credit quality (versus corporates). Agency MBS are all rated AAA, while the percentage of AAA in IG corporate and the Agg is 1.5% and 71%, respectively¹⁰ .

5. Lower correlations to equities: Unlike other yield-enhancing bond sectors, the higher credit quality leads to lower equity sensitivity and may also be able to mitigate equity risk similar to that of standard Treasuries—but without the duration risk. As shown here, the historical correlation of MBS to equities is essentially zero.

The impact of not including mortgages in a portfolio
The portfolio implications of not having a mortgage exposure can be witnessed by removing MBS from the broader Agg and examining the risk and return characteristics. Removing MBS creates a 50%-plus weight in Treasuries and 30%-plus in IG corporates, historically. The remainder goes to government-related exposures. Given the properties discussed earlier, the natural impact is intuitive. The yield declines, the duration is extended, and the volatility is increased. The charts below show these metrics on an average basis dating back to 1997.

From a cumulative perspective, the return from the Agg with no MBS does improve given how strong corporate credit has performed over the past year and the double-digit gains from longer-duration securities in 2019. However, on a rolling one-year basis, the Agg with MBS outperformed the Agg without MBS 79% of the time over 264 observable periods since 1997¹¹. And, the Agg with MBS had improved Sharpe Ratios, as shown below, with lesser drawdowns. The average 1 year return shown below is the average of the 264 rolling 1 year returns.

Portfolio construction perils can be avoided
From an implementation perspective, there are two options to obtain mortgage exposure: own pure beta to earn the yield premium and the volatility reduction MBS has displayed over the last 20-plus years, or find an active manager with a decided overweight to mortgages who can adeptly manage the two primary risks of MBS (prepayment or extension risk), structuring portfolios to mitigate unexpected shifts in the securities cash flows.

From a strategic perspective, not owning mortgage-backed securities or having a relative underweight portfolio can be a perilous portfolio construction decision. With more corporate bond exposure in portfolios identified, it may make sense to up the MBS exposure in strategic allocations. And tactically, given the low yields, extended duration, and tight credits, MBS may be an ideal overweight position, sourcing it from corporate bonds, to balance risk in the hunt for yield as discussed in our 2020 Outlook.