I recently addressed the significant drop in book value that one could expect on Capstead Mortgage (NYSE:CMO), essentially due to hedge ineffectiveness and a particularly strong sell-off in hybrid ARM valuations in this article. In this other article, I also looked into American Capital Agency Corp.'s (NASDAQ:AGNC) expected book value, dropping less than expected thanks to hedge effectiveness and the composition of their book.
As it turns out, CMO is not alone with a bad surprise potential. In this article, I show why one should also be expecting a significant drop in book value for Hatteras Financial (NYSE:HTS), in the order of 20%. HTS is currently trading around 23.15, while I would estimate Q2 book value around 22.75.
Integrating the analysis on HTS and CMO with an updated analysis of AGNC's end of Q2 book value leads to a paired trade idea: long AGNC, which should benefit from a positive surprise when the market realizes its hedges worked as expected, and short CMO and/or HTS, which I think are likely to catch the market unaware.
In these analyses, I should stress that the marks on longer resets hybrids and 15-years that compose HTS's book are more precise than the marks on shorter resets or seasoned hybrids that are in CMO's book. The market for generic 5/1 ARMs for example is liquid enough that there is not much uncertainty on marks. As a result, I would have even more confidence in the magnitude of the book value drop for HTS.
AGNC's updated book value estimate
Following the logic laid out in the above-mentioned article on AGNC, I reflected market prices, interest rate curves, swaption volatilities and prepayment behavior up to 6/28. The expected drop in book value has slightly increased, but remains in the same order of magnitude:
- Income from MBS or TBAs (equivalently assuming delivery) of $780m, from which we take out $167mm in swap negative carry, and $115m in premium loss due to prepayments and $78m in repo, for a net of $420m;
- A loss of $3.46bb on the MBS and TBA positions;
- A gain of $2.34bb on swaps, swaptions and bond shorts.
All that resulting in a net value added of negative $700mm. Then taking out $415m of dividend payments (which incidentally are very much in line with the real income after accounting for hedge carry), and about $55m of management fees, we have a total expected book value change of -1.174bb. This corresponds to $2.97/share, or a book value drop of 10%, to $26.97. The day after Q1 book value was announced on May 2nd, AGNC closed at $29.27, for a BV of 28.94, that is a premium of just 1%.
If the same premium were applied to AGNC's book value, and it is indeed down by 10% over Q2, its stock price would be $26.26, that is almost 20% higher than its current level. This appears to me as a long opportunity.
The hedges put on by HTS were purely composed of swaps, spanning a relatively small range of maturities, up to less than 5 years. As can be seen in the table below, the total notional of these swaps was $11bb, to compare with about $23bb in face value of MBS positions. The average duration on these swaps is negative 2.6, not particularly large given the product breakdown in HTS books, as we will see below. Without digging further it should already be clear that, at least from the end of Q1 positions, HTS's positions were under-hedged.
In addition to an issue with the amount of swaps, there was also an issue with the part of the curve they were hedging. As I pointed out in the article on CMO, referenced above, the short end of the yield curve did not move much in Q2, versus the long end. As a result, short swaps were generally almost entirely ineffective as hedging instruments, versus longer swaps.
This shows in the all-in projected performance of HTS's hedges: relative to notional amount, the longer swaps performed much better than the shorter swaps: for example $1.4bb of 4-yr and longer swaps made $28mm, versus $3bb of swaps under 2 years that only made $13mm. Including the negative carry on the swaps and mark-to-market value changes the projection is a gain of $57mm on the hedges.
Hatteras's MBS holdings
HTS's holdings can be summarized as follows: a large majority of hybrid ARMs spanning a wide range of times to reset, and some 15-year FRMs with low coupons (3s and 2.5s).
These are not considered "risky" MBS, for a variety of reasons, among others:
- Both asset types tend to exhibit prepayments that are less directional than typical 30-year FRMs. Hybrids prepay faster, but these prepayments stay fast, in general, even when rates back up, which is a positive characteristic. On 15-years, the generally smaller loan sizes, combined with the smaller amount of savings (all else equal) in a refi, leads to slower "top speeds."
- Both asset types have shorter durations than comparable 30-year FRMs, leading to less exposure to interest rates, due to faster speeds in the case of hybrids, and shorter maturities in the case of 15-years.
Now, while these types of bonds are more "comfortable" than low to mid coupon 30-year FRMs, they do have some duration. And if that duration is not hedged, the result is potentially worse than a better hedged 30-year MBS.
In addition, what took place in Q2 was not simply a back up in rates, but also a generalized and broad cheapening of mortgage assets in general. So given rates, given implied volatility, given mortgage rates, MBS just traded cheaper than what one would have expected. That risk was not hedged at all by HTS, according to its end Q1 positions. Such risks would have been mitigated by appropriate (i.e. long) swap hedges, and convexity positions such as long-dated swaptions, but that was not HTS's approach.
The table below shows some details on the main estimated/representative MBS positions for HTS, with prices over the period along with some risk characteristics.
The drop in valuations on most of these bonds reflects the effect of duration (generally accounting for 1-1.5 points of loss) and the effect of underperformance versus hedges, or OAS widening (generally accounting for 1.5-2.5 points).
In total, we estimate an income of $162m, from which negative swap carry ($34m), repo ($22m) and premium loss due to prepayments ($30m) are subtracted, for a net value of $76m. Now this is a little higher than the dividend distribution of about $69m, but do note that in principle HTS should have had more, and different swaps, in order to hedge its book better. So this net income is net of suboptimal hedge cost; better hedges would have been more expensive.
In terms of book value, we estimate the drop in MBS positions value at $645m. Adding income and the gains on hedges, we end up with an economic added value of negative $478m. Subtracting dividends that were paid (conservatively neglecting management fees) we have a net amount of negative $547m, that is $5.54/share, or a drop by 20% to $22.73/share.
Over the past few weeks, the stock market seems to have recognized that HTS had some risk exposure, and that its book value would be down. At the end of Q1 HTS's book value was 28.27, and when that was announced on 4/30, HTS's stock price traded around 26.5, hence at a 7% discount.
If we applied that same discount to the projected book value for Q2, we get a stock price of 21.15, a full 2 points lower than current levels. I view this ~10% potential drop in price as fairly likely, given the current market environment.
Given the significant discounts at which some other REITs trade (such as AGNC for example), one could also reasonably argue that either these other REITs should trade closer to book, or a REIT like HTS should see a deeper discount applied, further amplifying the expected price drop.
Finally, if I compare my perspective on HTS and on CMO, I think that there is more noise around the CMO estimate (at the core, due to seasoned hybrid MBS market liquidity), but that it has more potential to create a surprise. I perceive my projection on HTS to be effectively more accurate, but with less surprise potential.
There are many complex assumptions that go into the analysis laid out above, but I will point out the following risks, in addition to those that I had already pointed out in the two articles previously referenced:
- The market's perception of interest rate exposure being what it is, if rates backed up in the near term, AGNC would likely suffer more than CMO or HTS, notwithstanding their fundamental exposure
- My analysis relies on various sources for MBS prices, but on the less liquid products, prices are not clear-cut, and any REIT could use the "nuclear" option of declaring market prices were too stressed or not really usable, and instead provide their own marks (therefore potentially biased).
The current situation strikes me as an opportunity due to the market's lack of detailed analysis of a stock they believe is particularly risky and exposed , and stocks they believe are particularly safe (HTS and CMO).
Disclosure: I am long AGNC. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.