The Big Idea

A quick guide to the leveraged trade in agency MBS

Steven Abrahams | June 9, 2023

This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors. This material does not constitute research.

Put together the recent wide spreads and the leverage routinely available in agency MBS, and the rising interest in the leveraged trade makes sense. Plain pass-throughs currently show potential for double-digit returns on equity. And adding specified pools and other twists to the trade raises those numbers. But rarely does anything come for free. There is risk in the trade that warrants that kind of return.

To understand the risk and reward in leveraged agency MBS, it helps to walk through different elements of the trade. And there are several:

• Simple carry
• Hedged carry
• Option-adjusted carry
• Methods of adding to returns
• The significant basis risk in the trade, and
• The challenge of pinning down the volatility of its ROE

For most of the last decade, agency MBS has traded at spreads that left little room for error in a leveraged position. Current spreads look much more forgiving, although managing the position still remains a high art.

Simple carry: simple, but misleading

A lot of investors first look at the leveraged MBS trade through the simplest version of carry—the expected income or cash flow on the asset net of the expected expense of financing. That’s usually easy to calculate, but misleading. It leaves the impression that the trade, or any leveraged trade, depends largely on the difference between asset coupon and the cost of funds. The cost of funds plays a part, but not as big as you might think. An explanation.

Simple carry reflects all the differences between MBS and funding risk, and those differences fall into two distinct categories:

• Differences in the fair value of taking duration risk, with MBS offering compensation for longer duration than funding, and
• Differences in the fair value of prepayment or convexity risk, with MBS offering compensation for much more prepayment risk or negative convexity

If you momentarily ignore those differences, today’s deeply inverted yield curve and high rate of repo financing take a big chunk out of any MBS cash flow. Take a simple financed position in a 30-year 5.5% pass-through, for example (Exhibit 1). A typical MBS REIT might put in $12.5 million of its own cash and borrow $87.5 from the repo market to buy $100 million in market value MBS. The MBS yields an expected 5.48% and earns an expected $5.48 million in annualized interest. But with the repo market charging 5.13% to finance MBS, the interest expense is an annualized $4.49 million. That leaves less than $1 million in net income—a 7.93% return on the investor’s own cash—to take the combined duration and convexity risk of the funded position. Okay, but not compelling.

Exhibit 1: A current financed position in 30-year 5.5% pass-throughs

Note: All market levels as of COB Jun 2, 2023. BAM models of option-adjusted risk and spreads.
Source: Bloomberg, Santander US Capital Markets.

In fact, if you concentrate only on carry, you come away with the impression that the leveraged MBS trade depends mainly on the shape of the yield curve. In the hypothetical case where everything stays constant but repo drops back to the 6 bp rate before the Fed started hiking, net income would leap to $5.43 million and ROE to 43.42% (Exhibit 2). Compelling, but it is a mash-up of fair return to duration and convexity risk that, yes, can vary widely with the yield curve. That’s not a clean, leveraged trade in the unique risk of MBS.

Exhibit 2: A hypothetical financed position in 30-year 5.5% pass-throughs

Note: All market levels as of COB Jun 2, 2023 except for MBS repo, which is arbitrarily set to levels of March 2022. BAM models of option-adjusted risk and spreads.
Source: Bloomberg, Santander US Capital Markets.

Hedged carry: better, but not complete

By hedging out duration risk, the potential returns to taking prepayment risk or negative convexity—the distinguishing risk in MBS—becomes clear. Hedging out duration risk also makes clear that the key spread is not the one between asset coupon and cost of funds but rather between asset yield and the yield on the duration-neutral hedge. This means investors can find attractive leveraged positions in any asset with positive spread to the curve, including the lowest coupon in the pass-through stack.

The results from a leveraged, hedged position in MBS shows most easily by returning to the example of a financed position in a 30-year 5.5% pass-through. Most investors have a menu for hedging rate risk that ranges from shorting Treasury debt to using Treasury futures or swaps. Economically, all should land in the same neighborhood.

An investor could hedge rate risk in the pass-through by paying the fixed rate on a combination of 2- and 10-year SOFR swaps. In the case where the investor choses to reduce duration exposure to zero, the investor would size the positions so the negative duration in the swaps exactly offsets the positive duration in the financed MBS (Exhibit 3). In this example, the more than $3 million in dollar duration in the financed pass-through gets exactly offset by more than $3 million in dollar in the swaps.

Exhibit 3: A financed, duration-hedged position in 30-year 5.5% pass-throughs

Note: All market levels as of COB Jun 2, 2023 except for MBS repo, which is arbitrarily set to levels of March 2022. BAM models of option-adjusted risk and spreads.
Source: Bloomberg, Santander US Capital Markets.

This example also shows how the financing rate gets neutralized in a hedged position. It is important to note that in any rate hedge, the investor pays the fixed rate on one leg and receives the floating rate on the other leg. The combined income from the floating legs on properly sized swaps exactly offsets the cost of the repo financing for the MBS. In this example, the $1.11 million in floating income from the 10-year swap and the $3.37 million in floating from the 2-year swap combine to $4.49 million, offsetting the cost of MBS repo. Put another way, in a hedged and leveraged position in MBS, the repo rate—and the simple carry along with the shape of the yield curve—does not matter. The leveraged hedged position nets an expected $1.84 million or a 14.73% return on $12.5 million in equity. The repo rate could fall to zero or rise to 20%, just to pick two arbitrary numbers, and all else equal, the ROE would come in at 14.73%.

Option-adjusted carry: getting closer

In a duration-hedged position, an MBS investor still has negative convexity. As rates move higher, the duration of the MBS will get longer. To keep the net duration near zero, the investor will have to effectively add to the initial swap. As rates move lower, the duration of the MBS will get shorter. Now the investor will effectively need to buy back some of the initial swap. This rebalancing, usually called delta-hedging, can get expensive and reduce net income over time.

Instead of delta hedging, the portfolio could buy options to protect against rate moves. But the price of the options, as options pricing makes clear, should be almost exactly equal to the cost of delta-hedging. The net spread after delta-hedging the position is approximated by the option-adjusted spread on the MBS. Bloomberg’s model currently estimates that OAS on TBA 30-year 5.5% pass-throughs as 62 bp. That means delta-hedging would convert the pass-through to a position that floats at 1-month SOFR + 62 bp. That equates to a 10.09% ROE (Exhibit 4). If a static duration-hedged position starts at 14.73% ROE and a delta-hedged position ends at 10.09% ROE, the difference is the cost of constantly hedging risk.

Exhibit 4: Estimating ROE from OAS on a 30-year 5.5% pass-through

Note: Investors can also calculate ROE = OAS x Leverage + Funding Rate. This equation makes clear that the funding rate does play a role: the higher the funding rate, all else equal, the higher the ROE. Unlike a simple carry calculation, this makes clear that higher funding rates are good for ROE. All market levels as of June 2, 2023.
Source: Santander US Capital Markets.

Adding to returns

Of course, a portfolio does not have to constantly hedge all risk, and many portfolios do not. Some portfolios take risk because they have proprietary advantage in some area or another. Some take risk simply because they need to generate dividends to compete with other funds. A portfolio can add to returns in a range of ways that involve taking more net risk. For example:

• Reducing hedges and taking net duration risk
• Rebalancing less frequently and taking net duration and convexity risk
• Hedging with one point on the yield curve and taking key rate duration risk
• Using negative duration interest-only MBS to add carry and take net convexity and volatility risk

Of course, a portfolio can also add to leveraged and hedged MBS returns by selecting securities with better OAS:

• Across 30-, 20- or 15-year MBS
• Across Ginnie Mae or conventional MBS
• Across different coupons
• Across different specified pools

And an investor can add to returns by lowering the cost of financing. That might not be easy in the repo market, but the TBA dollar roll market often finances at implied rates well below repo. That can make a significant difference.

Finally, leverage can obviously have a dramatic effect on ROE. Every turn of leverage multiples every basis point of net asset spread. And agency MBS can get leverage in the repo market in amounts second only to US Treasury debt. Non-agency MBS or other forms of credit may have wider nominal or option-adjusted spreads than agency MBS, but leverage usually is much more limited and the potential ROE often falls short of the return available in agency product.

Just as leverage multiplies return, it also multiplies risk. That brings us to basis risk.

Still long basis risk

Even after hedging duration and convexity and ending up with the OAS on the asset, the position is still exposed to basis risk. And the risk is substantial. Remember that the portfolio owns MBS and hedges with swaps or Treasury debt or futures and related options. Based on shifts in supply or demand for MBS or views of prepayment risk in MBS, spreads on MBS can go wider or tighter even if nothing changes in the rates markets. Changes in spreads translate into changes in MBS price, and that flows directly into equity.

Back to the example of the leveraged TBA 30-year 5.5% pass-through:

• At $99-27 and at a median 212 PSA, the pass-through trades with an average life of 7.30 years and a spread of 173 bp
• If spreads widened 25 bp to 198 bp, the price would drop to $98-16
• If spreads tightened 25 bp to 148 bp, the price would rise to $101-06
• Over that 50 bp range in the spread basis, the price swing of $2-22 represents 21.5% of the total equity in the leveraged position. Put another way, over a 100 bp spread range, the position has a 43% spread duration of equity!

The basis risk in a leveraged MBS position hedged with rates—even delta-hedged—explains the ROE available. The potential double-digit returns reflect the relatively high duration of equity.

An investor can try to mitigate some of the basis risk by using interest-only MBS (IO) instead of rates hedges to offset the duration of a pass-through. An IO shows negative duration, rising in price as rates go up (and prepayments fall) and falling in price as rates go down (and prepayments rise). The investor can offset the positive duration of the pass-through with the negative duration of the IO. IO also tends to show negative spread duration, rising in prices as MBS spreads widen (since mortgage rates rise and prepayment fall) and falling price as MBS spreads tighten (since mortgage rates fall and prepayments rise). The caveat to using IO is that the market is much smaller and less liquid than the pass-through market, so this approach probably only suits smaller leveraged positions in MBS.

The final frontier: the Sharpe ratio of the position

With MBS spreads relatively wide, leverage readily available and potential returns attracting a lot of attention, it leaves open the question of the volatility of the ROE. An investor presumably has a range of alternatives, so the position’s Sharpe ratio or expected return-per-unit risk matters for allocating capital.

Since the biggest risk in many hedged positions is basis risk, the biggest influence on the volatility of ROE usually is the volatility of the basis. An investor can use the historic volatility of the basis as a starting point. But the only volatility that matters is the volatility that plays out in the future while the investor holds the position, and that future volatility is hard to gauge. There is one place, however, that offers at least some market-implied view of future mortgage basis volatility: the market for options on the equity of mortgage REITs.

The equity value of REITs that predominately invest in agency MBS and hedge with rates depends materially on shifts in the mortgage basis. That’s not the only important influence, of course. The starting spreads on acquired assets, net duration and convexity risk, the changing amount of leverage, dividends in competing REITs and other things have a big impact, too. But basis risk is important.

One benchmark for the volatility of equity in an agency MBS REIT is the market for options on AGNC, a REIT that primarily invests in agency MBS and hedges with rates. Other REITs could serve as benchmarks, too. That market provides implied volatility based on 1-year options struck at the daily price of an AGNC share. Over the last five years, that implied volatility clearly shows a pre-pandemic regime with relatively low implied volatility and a post-pandemic regime with much higher implied volatility (Exhibit 5). Options imply the post-pandemic risk in a leveraged and hedged agency MBS portfolio is higher.

Exhibit 5: Implied volatility of equity in a leveraged, hedge agency MBS portfolio

Note: The data show Black Scholes volatility based on rolling 1-year options struck daily at the price of AGNC, a mortgage REIT.
Source: Bloomberg, Santander US Capital Markets.

The rising implied volatility of equity in a leveraged and hedged agency MBS portfolio raises the question of whether current OAS offers enough compensation, at least relative to history. There is room for much more careful analysis of this, but some quick analysis suggests current OAS at least offers fair compensation (Exhibit 6). Current implied vol is 27% and par 30-year OAS is 52 bp, suggesting just under 2 bp of compensation for every 1% of equity risk. In 2021, implied vol averaged 19% with OAS at an average -12 bp, leaving MBS looking rich relative to risk. In 2019, implied vol averaged 14% with OAS at an average of 28 bp, again suggesting 2 bp of compensation for every 1% of equity risk. Current OAS relative to expected basis volatility looks fair to 2019 levels.

Exhibit 6: Option-adjusted spread to Treasury debt on a par 30-year MBS

Note: The data show OAS on par 30-year conventional MBS using Bloomberg models.
Source: Bloomberg, Santander US Capital Markets.

Spreads and leverage draw new interest

Inquiries about the leveraged trade in MBS have definitely picked up since March as nominal spreads and OAS have widened. Returns on the trade from here should depend, as it almost always does, on the direction and volatility of the MBS basis. A simple look at the options market implies that the trade is fair value. But with good security selection, proper hedging and leverage and anything that improves the cost of funds, it could turn out much better than that.

* * *

The view in rates

OIS forward rates now put nearly a 70% probability on a 25 bp Fed hike by August followed by a 25 bp cut by December. For practical purposes, this looks like a Fed on hold. After the FOMC meeting on June 14, the market will likely turn to figuring out how long the Fed will hold policy steady, and that will influence fair value all along the yield curve.

Other key rate levels:

• Fed RRP balances closed Friday near $2.13 trillion, the lowest level since mid-March. Attractive levels on T-bills in the wake of the debt ceiling resolution is almost certainly drawing away RRP cash.

.

• Settings on 3-month LIBOR have closed Friday at 554 bp, up 4 bp from a week ago. LIBOR goes away after this month. Setting on 3-month term SOFR closed Friday at 524 bp, up 1 bp from a week ago. The gap between LIBOR and SOFR is wider than the recommended ARRC spread of 26.16 bp
• Further out the curve, the 2-year note traded Friday at 4.60%. With the Fed likely to hike again and hold fed funds around 5.375% into next year, fair value on the 2-year note is around current yields. The 10-year note closed at 3.74%. With inflation likely to drift down and growth likely to slow, fair value on the 10-year note is lower than current yields.
• The Treasury yield curve traded Friday afternoon with 2s10s at -85, flatter by 5 bp over the last week. Expect 2s10s to flatten further. The 5s30s traded Friday morning at -3 bp, 7 bp flatter over the last week.
• Breakeven 10-year inflation finished the week at 221 bp, nearly unchanged on the week. The 10-year real rate finished the week at 155 bp, higher by 4 bp in the last week.

The view in spreads

With the debt ceiling resolution, volatility has dropped off. The June 14 FOMC should lower it further. That should help nominal spreads in MBS and credit. The Bloomberg investment grade cash corporate bond index OAS closed Friday at 163 bp, wider by 5 bp in the last week. Nominal par 30-year MBS spreads to the blend of 5- and 10-year Treasury yields traded Friday at 164 bp, down 12 bp in the last week. Par 30-year MBS TOAS closed Thursday at 51 bp, down 9 bp on the week.

The view in credit

Most investment grade corporate and most consumer balance sheets look relatively well protected against the likely impact of Fed tightening. Fixed-rate funding largely blunts the impact of higher rates on both those corporate and consumer balance sheets, and healthy stocks of cash and liquid assets allow these balance sheets to absorb a moderate squeeze on income. But other parts of the market funded with floating debt look vulnerable. Leveraged and middle market balance sheets are vulnerable, especially with the sharp tightening of bank credit in the wake of SVB. Commercial office real estate looks weak along with its mortgage debt. As for the consumer, subprime auto borrowers, among others, are starting to show some cracks with delinquencies rising quickly. The resumption of payments on government student loans in September should add to consumer credit pressure.