By now most market participants are resigned to the fact that short term rates WILL be low for “an extended period” of time.  Maybe you have been fooled into thinking QE was/is inflationary, and possibly even acted on a popular yet ill-conceived plan to buy TBT thinking that rates MUST go up.  Similarly, for those stuck in the ever present state of fear mongering over the debt ceiling, a lesson from Cullen at PragCap may be in order.

This past weekend, good friend Howard Lindzon noted the genius of the banking system where “The banks borrow money from the government at a low rate and lend money at a higher rate…to the government.  That’s genius”  

What I will propose will take advantage of depressed short term rates and little known sections of the fixed income markets, a formula that yields a tremendous risk/reward opportunity across bull or bear markets.

Barron’s Has it Wrong:  In this weekend’s Barron’s, columnist Dimitra DeFotis lays out “Two REIT’s with double digit dividend yields”  To her credit she is sniffing around in the right place, however breakdowns in her analysis actually point readers towards two mortgage REITS that I believe actually have greater amounts of risk and poorer risk/return profiles than alternatives.

Various mortgage REITs hold both Agency & Non-Agency MBS.  Understanding the risk/return profile of each across positive & negative macro-economic scenarios is ESSENTIAL in being able to ascertain the relative attractiveness of each REIT.

Agency MBS- 

1. For those unfamiliar, Agency MBS represents Fannie Mae, Freddie Mac, and Ginnie Mae securitized mortgage pools.  Sometimes REITs will purchase CMO’s, but we will assume the majority of purchases are passthroughs, where each prepayment gets distributed to the bondholder on a pro-rata basis. 
2. Mortgages that meet certain standards can sold to Fannie/Freddie (GSE’s) and put into an Agency MBS.  On the run 30yr pools are among the most liquidly traded products on the market.  15yr pools are also common.  MBS are trading at record high prices as QE has flooded the market with liquidity and has caused all risk assets to be bid up.  30yr 4’s (30year 4% coupon) trade at ~101 cents on the dollar.  Higher coupons obviously trade at higher premiums.  30yr 6’s trade close to 110 cents on the dollar.
3. The drawbacks of Agency MBS are obvious.  Every time a prepayment takesplace, the bondholder receives his premium bond back at par - thus a loss. Economically, it is not really a loss as nobody buys a 30yr 6 expecting a 6%, but nonetheless you can see the issue with prepayments.  
4. They very best case for Agency MBS is that rates continue to rally and prices rise. As is with the case of MBS, they are negatively convex, so at the precise time when you want the bond to be longer (when rates go down), homeowners prepay.  Conversely when rates rise, you would like to re-invest your money in something with a higher rate, but borrowers are in no hurry to prepay a mortgage at a below market rate.
5. Assuming no leverage, what will the return profile look like across interest rate scenarios if we purchased a fixed rate mortgage.  The scenarios we will look at is Rates down 100, Flat, Up 100, Up 200, and Up 300 ramped across a 1 year horizon.  When we measure return, we will look at TOTAL RETURN, that is income plus/minus price appreciation.  Listed below is a total return analysis for a 15yr 3.5% MBS.

What you are looking at above is a total return analysis of this particular 15yr 3.5% security under various rate scenarios over a 1yr horizon.  Best case scenario under down 100 (see MTGE column on the right) would be a 3.6% total return.  A rates up 300 would mean a NEGATIVE 8% return.

Due to the agency backing, REITs that buy a lot of Agency MBS are able to obtain greater amounts of leverage.  So if they would obtain their 15% yield by borrowing at 1% (just an example), and buying 15yr 3.5% at 3%.  From simple math you can see they need to lever up that 2% spread significantly.

Non-Agency MBS

Non-Agency MBS have no government backing of losses, instead they were set up with credit support, i.e. a greater amount of loan $’s than amount due.  This padding would insulate bondholders from potential losses.  What I will try to articulate in this example is that the risk of negative returns from owning a non-agency MBS bond is very low.  The interesting part is that despite the negative connotations with RMBS (toxic bonds etc…), due the discount nature of the bond and higher income, they have a far superior risk/return at almost every interest rate/economic scenario.

BAFC 2006-F 1A1

As of year end this bond had a par of ~$23.1mil and a large fund was marking it at $20.2mil or ~87cents on the dollar.  Fitch has it rated CCC and S&P BBB.  As a rough rule of thumb, if a bond is projected by a rating agency to take at least 1 dollar of loss, it is automatically a CCC.  That doesn’t tell you whether they expect 1% loss or 100%, just SOME loss.

The underlying collateral is 5/1 hybrid ARMs with a weighted average reset date of December 2011.  They will reset to 2.75% above the 1yr treasury.  With the current coupon of 3.19%, we can expect further declines in the months ahead.

Unlike Fannie and Freddie pools where principal & interest is passed through to mortgage holder (hence called a pass-through), many private mbs bonds are structured to provide cash flows to different classes.  This particular bond is a super-senior (group 1a1) with the (1a2) as a support bond.  So if/when losses hit, they will first be taken by the 1a2 class.  As you can see below, our 1a1 bond currently has 10.3% credit support.  So the first 10.3% of losses won’t hit our tranche.  The B classes below are the supports for all the senior classes (1, 2 , and 3) but they are almost completely wiped out.  

For the purposes of this basic high level analysis, just remember that the senior supports are supporting the super senior classes such as the relationship between the 1a1 and 1a2 class.

The next thing you are probably asking is what kind of loans are in the bond.  Most of you have heard these type of bonds referred to as “toxic” etc.   In my opinion whether or not it is toxic depends on the price you paid.  If you paid par ($100) and you only get back $50, that is obviously painful.  But it’s a great deal if you paid $50 and got back even $60.  Many institutions such as banks cannot hold these type of bonds due to ratings, capital etc.   

Looking at the delinquency history you can see that as of April 2011, 17.2% of loans from group 1 were in the 60+ day delinquency bucket.  This sums up lines 4-8.

The average person would view this and tell me that the 60+ delinquencies have increased since October from 15.8% to 17.2%.   That is true, and the obvious concern is that they will continue to deteriorate.   What about the rest of the loans? Looking at past payment history is a reasonable predictor of the future.  Homeowners who have fallen behind on their mortgage and are now current (reperformers) have a greater likelihood of being non-performers again.  Looking at group 1, over 75% of the loans have been current for at least 24 straight months.

So let’s recap.  The 1A1 bond has 10.3% of credit support and 17.2% in delinquencies.  If 100% losses were taken on all delinquent loans, the loss to our bond would be ~7%  (17.2-10.3).  Loss severity is the percentage of loss given a default.  So if a $100k mortgage goes into foreclosure and $70k net is recovered, the loss severity is 30%.  Historical loss severities over the life of this bond have averaged around 32%.  Meaning if the 17.2% of delinquent loans were all liquidated tomorrow with a 35% loss rate, then 5.5% of losses would be taken.    Remember, the bond has 10.3% of Credit Support, so no loss would be taken to our specific 1A1 tranche.

Let’s take our projections a step further, we will assume that 35% of all loans will default (Recall 75% of all loans have been current for at least 2yrs straight).  35% x 32% severity = 11.2%.  This is about the break-even point.  After taking away our 10.3% CE, our particular bond would take a ~1% loss.  This is admittedly a “back of the envelope” way of projecting losses. Investors typically run prepayment and default projections through Intex to come up with a detailed set of cash flows.  The goal is to illustrate what has to happen for the bond to take a loss.

Remember, this bond is valued at 87 cents on the dollar in the market, so there is the potential upside of 13 cents of accretion.  For each dollar of prepayments that comes in, you are receiving $1.00 paid at par for a security being held at $0.87.  That part of the beauty of buying discounted bonds.

Taking a step back and looking at the macro picture, there are a bunch of reasons why an investor would want to include this type of asset class into their portfolio.  First, the lions share of the bonds are either hybrid arms or floaters which will perform favorable in an increasing interest rate environment.  Secondly, an economic recovery would bode well for the more distressed holdings as firmer home prices and increased jobs would take pressure off of defaults and loss severities giving the potential for greater principal recoveries.   While RMBS is not at the insanely cheap levels such as when I wrote an article on TSI , they arguably have a much better risk return profile than high yield.  Losses are already built into the prices of these non-agency MBS bonds while high yield is reflecting an environment with significantly lower than historical defaults.

The Opportunity:  Buy Two Harbors Mortgage REIT ($TWO)

Unlike many of the other mortgage REITs, Two Harbors has a large percentage of non-agency MBS which is NOT interest rate sensitive.  In fact, as I said above,  you can argue that the bonds will IMPROVE if rates go up.  Summary:

1. Yield:  When you buy Non-Agency MBS, you are buying at loss adjusted yields, meaning you start with the assumption that things DO NOT improve in the housing market.  So even if things don’t improve, loss-adjusted yield are in the 5% range.  If they do improve, yields go up to the high single digits, low teens.
2. Funding: Like other REITS, TWO is able to borrow based on short term rates (mainly LIBOR).   However, the Agency focused REITS need to lever up much higher to obtain the yields that they do.  The leverage of TWO is much lower at 3-4x.  They are still able to obtain a mid teen’s yield as their RMBS holdings have such a higher yield.
3. Hedging:  Two Harbors holds plain vanilla derivatives to hedge to potential of rising interest rate risk.  They currently project that a 100 basis point rise in rates would equate to a 2% decline in economic value. 
4. Rate Risk: As we have discussed ad nauseum, non-agency is NOT interest rate sensitive, thus the rate risk is significantly less in Two Harbors.  Meanwhile, agency MBS is highly rate sensitive.  Even ARMs would get hit hard in a rapidly rising rate environment due to caps.

Conclusion:  The earnings power/discount nature of Two Harbors’ assets provide a huge margin of safety when dealing with credit & interest rate risk.  With lower leverage than other REITS and less interest rate sensitivity, it’s risk/return profile should leave investors much more comfortable regardless of the rate scenario.  Trading at ~$10.25 and a quarterly $0.40 dividend, rates need to increase substantially to offset the massive amount of income earned.  Plays such as $TWO are prime for this period of “extended low rates”.