Description

Two Harbors Series C Preferred

DISCLAIMER: Don’t be dumb - I’m just a random person on the internet whom you’ve likely never met. If you trust this write-up, you deserve the losses you will reap and your boss and clients should fire you...but…maybe? 

—----------

Opportunity:

Two Harbors Series C preferred offers a reasonably safe and mathematical multi-year teens IRR.

mREIT Preferred Overview:

As you may know, most mortgage REITs (mREITs) issue preferred equity securities. These financing tools are targeted at retail investors to add juice to the ROE of the common equity. Preferreds are used primarily as a regulatory and/or tax arbitrage - yieldy products that pay dividends instead of interest and are treated like equity (but are senior to common). I’m simplifying for brevity, but preferreds typically have a par value of $25 (such that a pref trading at $20 is trading at 80% of par). Most issuances are quite small and trade like microcaps.

Most mREIT preferreds are perpetual, but can be redeemed (called) after a certain period of time. The general flavor of an mREIT preferred is that it offers a fixed coupon (dividend) for an initial number of years (e.g., 5 to 7 years). After the initial fixed period ends, most mREIT preferreds switch from a fixed coupon to a floating coupon (e.g., SOFR + 450bps) and they become redeemable from that point forward. 

At the time most outstanding preferreds were issued, ZIRP prevailed and the floating coupon was expected to be similar to or less than the fixed coupon. The world has changed.

Example:

I will come to Two Harbors in a moment. Before I do, I want to highlight three preferreds issued by a different mortgage REIT: AGNC (ticker: AGNC). As you may know, AGNC is one of the largest mortgage REITs in the world. Like most mREITs, it has a variety of preferred equity securities outstanding - currently five issuances (Series C, D, E, F, and a newly issued G). 

Series D and C have been around for several years - let’s touch on those first.

AGNC Series D: In February 2019, AGNC issued a 6 ⅞ Series D preferred. Its fixed dividend of 6 ⅞% lasts until just over one year from now, on April 15, 2024. At that point, it switches to 3-month LIBOR + 4.83%. As of this writing, that would be a 9.2% par coupon. That Series D preferred trades for $22.43 (90% of par). That means the 6 ⅞ coupon yields 7.7% during the remaining fixed period and the 9.2% (at par) floater would yield 10.3% (at market), if it were floating now.

AGNC Series C: Similar to the Series D, AGNC has a Series C preferred. The Series C was issued in August 2017 at 7.0% fixed. Its fixed period recently expired, making it redeemable and its dividend (coupon) perpetually floating. The floating coupon is 3-month LIBOR + 511bps. With 3M LIBOR at 4.83% as of this writing, the floater is a 10.0% coupon at par. The Series C trades for $25.33 (a slight premium to par). This is key: the pure-floating Series C trades over par and yields 9.9% while the fixed Series D trades at a discount to par and its implied floating yield is 10.0%

AGNC Series G: The Series G was issued a few months ago in September. It is a 7.75% that resets October 2027 at the Constant Maturity 5-Year US Treasury (H15T5Y Index on Bberg) + 439bps (or ~8.4% at par). This is trading at $23.47 or 94% of par, yielding 8.3% now. That 8.3% is probably a fair indication of where AGNC could price a new preferred today. (Note: the Series G is a bit unusual in that it doesn’t truly “float” in 2027, but instead “resets” to a new fixed rate every five years based on the then prevailing 5-Year UST yield + 439bps)

Below is a summary of these three AGNC preferreds:

Perhaps you see where this is going - in the current rate environment, the moment the fixed flips to floating, the coupon becomes very valuable to the owner of the preferred. Likewise, in today’s rate environment, the floater is very expensive for the issuer. You can see that the new Series G would cost ~8.3% to issue, yet the Series C earns “extra” yield of ~1.6%.  Why is this?

It is because the Series C is callable - AGNC can take it out at par at any time. Given its nearly 10% cost to AGNC, the market assumes it will be called, which prevents it from trading substantially below or above $25. And AGNC hinted that this is coming when it issued the Series G:

Key Takeaway: 

Once they float, old prefs are expensive. However, once they float, they also become redeemable - this drives a migration toward par if they can be replaced at a lower coupon. One can imagine that has implications for the Series D pref above: if rates stay in the current ballpark, a buyer today would make a 7.7% fixed annual coupon for the next 14 months, then flip to a 9.9% floating coupon thereafter. Because that is expensive, ceteris parabis, you’d also capture the pull to par of ~12%. That creates a 14-month return of ~20% in a fairly safe security, after which you get paid 10% to wait. 

Setting aside a crisis, the risk to this is if the credit spread cost of new issuance rises substantially. At 433 bps over LIBOR, if AGNC has to issue at 600 bps over benchmark rates, then the Series D would probably “only” return to that spread. 

The ideal is to find a security that optimizes for both high yield and value generated by a pull-to-par in order to maximize potential total return. 

Two Harbors Preferreds:

Now, let’s talk Two Harbors. 

Two Harbors is a straightforward Agency-focused mREIT (NYSE: TWO) with three preferreds outstanding: Series A, B, and C. Series A and B are fixed (and non-callable) until 2027. However, C is callable in less than two years (January 2025).

The Series C has a fixed coupon of 7.25% and a floating coupon of 3-month LIBOR + 501bps (~9.9%). At $21.06, it yields 8.6% currently (and would yield 11.7%, if the floating period began today). We don’t know precisely where Two Harbors would be able to issue new preferreds today, but the Series A and Series B have to 4.25 to 4.5 years, respectively, before they flip to floating. In this sense, they are not too dissimilar from a new 5-year fixed preferred. Each has a fixed yield around 9% today and implied floating yields near the Series C. 

If we assume that the Series C floating yield would be ~10% if the floating began today, then, all else equal, the Series C will reprice to about $24.60 by that time, a 17% increase from today. Spread over two years, the price increase would be ~8.5% annualized and the fixed coupon will earn another 8.6% p.a. That combines to a 17% annualized return over two years, at which point you’d flip to floating and begin earning 11.7% (if interest rates in two years are similar to today).

Those are high yields for reasonably well-protected securities (see the three TWO preferreds below):

In the meantime, the preferred has optionality:

First, Two Harbors disclosed in its Q3 earnings report that, after quarter-end, it repurchased nearly 3  million preferred shares. It bought across all three preferreds, but focused on B and C, given they traded at the biggest discounts (428,549 Series A, 786,846 Series B, and 1,742,555 Series C). Two Harbors also just issued another $170 million of common equity in early February, adding more firepower to its balance sheet.

Second, preferreds that are nearing their floating period benefit from optionality in both directions. Obviously, if rates rise, floaters benefit (the shorter the remaining fixed period, the more favorable for rising rates). If rates fall, because the spreads on the floating period are fairly wide (501 bps, for Two Harbors Series C), at some level, it becomes attractive to refinance these with new preferreds, locking in the fixed rate period at then-prevailing lower levels (and paying you off at par).

When the floating period arrives, if the current rate environment prevails, Two Harbors is likely to seek to retire the Series C. Let’s talk about why.

mREIT Math:

An overly simple mortgage REIT trade today is for a given mREIT to use repo funding to lever its mortgage securities 5x. For example, today’s Fannie pools (yielding in the 5.3% range) can be financed with ~4.0% repo (many mREITs have legacy repo programs that charge lower rates but ultimately those will mature and be replaced with more current repo costs).

At 5x, that generates a straight return of:

5.3% + 5 x (5.3% - 4.0%) = 11.8% (before REIT SG&A)

Subtract 3% for REIT expenses = 8.8%

To take that 8.8% return and juice it, mREITs use preferreds as a form of leverage. With half a turn of our Series C preferred costing 7.25%, the yield levers to:

8.8% + 0.5 x (8.8% - 7.25%) = 9.6% ROCE

However, when that Series C resets to floating with a 10% cost, the leverage math doesn’t work. As a result, it is in the interest of the mREIT to redeem it and de-lever or replace with a lower-cost financing alternative.

I expect the Series C to be redeemed at par in roughly two years, providing a reasonably safe and income-centric high-teens two-year IRR.

Risks:

• Two Harbors is focused on Agency mortgages which are implicitly or explicitly guaranteed by the US Treasury. However, those mortgages are subject to prepayment risk, which becomes more acute in a falling rate environment. During peak-Covid fear, rates fell so fast that mREITs experienced substantial book losses and their preferreds plummeted in value as subordination buffers eroded

• Likewise, when rates rise, existing long-duration mortgages “extend” in duration, as prepayments plummet. While higher rates are bad for existing mortgage securities, they are attractive for redeploying capital into new, higher-yielding securities (either new mortgages or old MBS that have repriced)

  • It’s worth noting that Two Harbors runs a somewhat “hedged” portfolio by owning MSRs. Mortgage Servicing Rights receive a fee to service mortgages and therefore the longer the mortgage principal remains outstanding, the longer you get paid. The result is MSRs tend to make money when rates rise (and lose when they decline), providing a natural hedge to rising rates risk.

• Rates plummet, Two Harbors balance sheet takes a bullet, and the preferred’s floating rate benefit declines.

• These are small and illiquid
• mREITs are levered enterprises - leverage creates risk

NOTE: unlike bonds, preferreds accrue their dividends into the price of the security. For simplicity, I’m not backing out accrued dividends anywhere in this write-up. If I did, the prices of existing securities would be slightly lower and yields slightly higher. All the preferreds in this write-up pay quarterly dividends and the dividends are cumulative (i.e., if the mREIT stops paying them for any reason, they continue to accrue and must be made current before common equity can receive dividends).

Catalyst

• Rates continue to rise, making the floating period more expensive (and increasing the likelihood TWO redeems the Series C)

• Collect divvies

• Time passes and we pull toward par

• Collecting 9-10% p.a. heals many wounds