In this article, I look into how QE3, more particularly the Fed's MBS purchase program, affects agency mortgage REITs. I discussed my approach for agency mortgage REIT analysis in this prior article.

A few days ago, the Fed laid out its plan to buy about $40bb of agency MBS per month over the next several months. The price of mortgage-backed securities has since gone up, and their yields have come down significantly. As a consequence, many fear that REITs investing in these bonds would see their income drop, which would imply lower dividends in the future.

However, I think that in fact QE3 will not have a negative effect for two main reasons:

- The spread between primary and secondary mortgage rates will widen as the Fed pushes MBS yields tighter.
- The increase in book value of REIT MBS positions will largely compensate for the loss in interest.

I illustrate my logic with **Armour Residential **(NYSE:ARR). I like it as an example because I own it.

## Primary/secondary spread

An important metric in the agency mortgage market is the current-coupon. This is the net MBS coupon that trades at par. For example, in late October 2011, with FNMA 3.5s (30-yr FNMA generic MBS paying a net coupon of 3.5%) trading around 100.5, the current-coupon was about 3.31%. Around the same time, the 30-year mortgage rate as reported by the MBA was around 4.25%. The MBS current-coupon (the secondary rate) shows where mortgages trade as liquid securities, while the mortgage rate (the primary rate), as surveyed by the MBA, is the one you're paying as a borrower. The difference, the primary/secondary spread, which was 0.94% about a year ago, pays for origination, servicing, fees for guaranteeing the MBS, and a little excess spread for the issuing bank.

The key element here is that the secondary rate is indicative of the value of MBS, but the primary rate drives prepayments, since that is what borrowers see.

Now let's fast forward to today. The MBS market has priced-in the Fed's purchase program and yields have tightened. FNMA 3.5s now trade around 106.5. With no coupon trading below par, determining the current-coupon is a bit of an extrapolation. We can consider that a year ago, the coupon trading at 106.5 would have been about 4.85%. From that, we can infer that today's current-coupon would be at a historic low of 1.91%. At the same time, the latest MBA survey rate came at 3.56%. Hence the primary/secondary spread has massively widened to 165bps. This means that while MBS prices are going up, it does not translate in nearly as much of an increase in refinancing incentives.

In fact, only half of the rally in MBS yields has been transferred to the mortgage rates that borrowers see. This is extremely important because it implies that prepayments will not increase as much as one would otherwise expect. When rates rally because of a broad-based change in fundamental expectations, one anticipates that rates would move in parallel to a greater extent. But in the situation at hand, there is a particular distortion in that the purchases are disproportionately concentrated on "on-the-run" MBS.

One of the core reasons for the "stickiness" of the primary mortgage rate (not dropping as much as the secondary rate) are that mortgage originators just cannot ramp up their operations in a matter of days or weeks. It takes them a lot longer. In addition, the spread tends to be wider when volatility is high, as would be the case after the rate moves we have observed. So the primary/secondary spread will likely contract over time, but this will either require that rates back up, or that they stay stable, and will take at least a year. In the meantime, *prepayment speeds will increase but will remain relatively muted*.

## Limited reduction in REIT income

As the Fed buys MBS, their prices increase to a greater extent than prices on equivalent Treasury bonds or swaps. In other words, their option-adjusted spreads (OAS, see the article mentioned earlier for an explanation of OAS and some sources to track them) have tightened. As a matter of facts, OASs rallied by about 50bps on many MBS coupons. Hence this is not like a typical movement in rates, where OASs do not necessarily move much. If MBS go up, but Treasurys as well, then a REIT's position which is basically long the former and short the latter does not move much. But in the current situation, the *hedged* value of MBS is going up significantly.

Looking at ARR's position (detailed in this presentation for example), we can see that the MBS on the asset side have a duration of 2 years. If we assume that the bonds they own rallied by 50bps in OAS, this will correspond to an increase in net portfolio value of about 1%. Relative to equity, and with a leverage close to 10, it amounts to an increase by 10% in book value.

On the negative side, however, we need to reflect the fact that the yields at which ARR would be able to invest are going to be lower than previously.

But first of all, how much do they actually need to reinvest? Recent prepayments on their book came at 13% CPR. This is combining all kind of mortgages, so it is not very accurate, but should give us a reasonable approximation. We could assume prepayments rise by 50% in relative terms to 20% CPR, a moderate increase helped by the primary/secondary effect explained earlier.

Second, how much lower would yields be in this new environment versus what we had before the Fed's announcement? The proper way to measure that is net of various hedging costs; in other words, we should look at how much lower OASs are. As mentioned earlier, they are about 50bps tighter.

So we are looking at a total reduction in carry by 50bps x 20% of the balance per year = 10bps per year. What was the recent net spread on ARR's assets? In the above-mentioned presentation, I found that the assets paid 3.55%, but hedges (including duration and some convexity hedges) cost 1.18% on 8.5bb of swaps and 2.04% on 1bb of swaptions. In addition, there is 0.43% to pay in repos. All in, I find a net interest margin of 1.89%. Hence the true cost of a lower MBS yield environment is 10bps a year out of 189bps.

Hence, I would summarize the Fed's impact as follows for ARR, in my estimation:

- Book value up 10%.
- Future net margin declining by about 6% a year in relative term.

So as far as it does not go on for too long, this analysis implies that ARR should be OK.

Given my personal expectation that this round of QE will not need to go on for more than a year, I am not worried by its impact on agency REITs. The stock market's harsh reaction, as can be seen on REIT prices today, will eventually even out in my opinion.

Comments()